# Cloud Governance Tools: every page as markdown Preventive guardrails for AWS, Azure and Google Cloud: 100 documented entries with what each blocks and breaks, plus platforms ranked on safe rollout. --- # Cloud governance tools and the 100 preventive guardrails they put into production, with what each blocks and breaks Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 100 preventive guardrails: 40 for AWS, 30 for Azure and 30 for Google Cloud, each copied from provider documentation or a licensed open source set, with what it blocks and what it breaks. Act Security leads the platforms at 5.9, ahead of Sonrai Security at 5.7. - Identity and privilege: 14 - Data exfiltration: 8 - Public exposure: 10 - Logging and monitoring: 8 - Encryption and keys: 13 - Network perimeter: 7 - Region and residency: 5 - Organization integrity: 9 - AI services: 19 - Resource and cost: 7 Scope: security governance, the guardrail layer. Cost governance is out. ## The library by risk class, edition 10 Nov 2026 | Risk class | AWS | Azure | Google Cloud | All | | --- | --- | --- | --- | --- | | [Identity and privilege](/risk/identity-and-privilege/) | 7 | 3 | 4 | 14 | | [Data exfiltration](/risk/data-exfiltration/) | 2 | 1 | 5 | 8 | | [Public exposure](/risk/public-exposure/) | 4 | 3 | 3 | 10 | | [Logging and monitoring](/risk/logging-and-monitoring/) | 4 | 2 | 2 | 8 | | [Encryption and keys](/risk/encryption-and-keys/) | 7 | 3 | 3 | 13 | | [Network perimeter](/risk/network-perimeter/) | 2 | 3 | 2 | 7 | | [Region and residency](/risk/region-and-residency/) | 2 | 2 | 1 | 5 | | [Organization integrity](/risk/organization-integrity/) | 3 | 3 | 3 | 9 | | [AI services](/risk/ai-services/) | 7 | 7 | 5 | 19 | | [Resource and cost](/risk/resource-and-cost/) | 2 | 3 | 2 | 7 | 100 guardrails. Every entry gives the policy text, what it blocks, what it is known to break and how the provider says to test it first. - [AWS: 40 guardrails](/library/aws/): Service control policies (SCPs), Resource control policies (RCPs), Declarative policies, IAM permissions boundaries - [Azure: 30 guardrails](/library/azure/): Azure Policy (built-in definitions), Azure RBAC deny assignments (created by deployment stacks), Azure resource locks, Management group assignment - [Google Cloud: 30 guardrails](/library/gcp/): Organization Policy Service (managed and legacy constraints), IAM deny policies, VPC Service Controls (service perimeters), Custom organization policy constraints, Principal access boundary (PAB) policies ## Platforms by how safely they put a guardrail into production | Rank | Platform | Total | Removes (15) | Safe (30) | Class (30) | Fixed (10) | Id+Net+AI (2) | Native (10) | Audit (3) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.9 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.7 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.2 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.1 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 5 | [Turbot](/platforms/turbot/) | 3.9 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 6 | [Apono](/platforms/apono/) | 3.6 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.0 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8= | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.9 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 8= | [Wiz](/platforms/wiz/) | 2.9 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 10= | [P0 Security](/platforms/p0/) | 2.3 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 10= | [CrowdStrike](/platforms/crowdstrike/) | 2.3 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 10= | [Tenable Cloud Security](/platforms/tenable/) | 2.3 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 13= | [Entitle](/platforms/entitle/) | 2.0 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 13= | [Sysdig](/platforms/sysdig/) | 2.0 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 1.7 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.1 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | One change for a class of risk 30, safe to enforce 30, removes access 15, stays fixed 10, uses the cloud's own controls 10, evidence for auditors 3, identity, network and AI access 2. Scores 0 to 10 from the evidence read 10 Nov 2026; "=" marks a joint rank. Act is first at 5.9, ahead of Sonrai Security at 5.7. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on safe to enforce rose by 1 point (from 5 to 6) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. ## The platforms that put a guardrail into production most safely ### Rank 1: Act Security (5.9) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 2: Sonrai Security (5.7) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 3: AWS Control Tower (4.2) Provider guardrail service, One change, a class of risk 6 Highest score: one change, a class of risk, 6 of 10. Lowest: safe to enforce, 1. [Read the profile](/platforms/aws-control-tower/) ### Rank 4: Google Cloud IAM recommender (4.1) Provider tool, Safe to enforce 6 Highest score: safe to enforce, 6 of 10. Lowest: stays fixed, 2. [Read the profile](/platforms/google-iam-recommender/) ## Where each is strongest - **Act Security:** Act removes broad classes of access by simulating guardrail and ring-fence policies against a digital twin and delivering them as pull requests to the customer's IaC using cloud-native controls. - **Sonrai Security:** Sonrai blocks unused privileged permissions, dormant identities, third-party access and unused services and regions through org-level cloud-native policies, and restores access through ChatOps approvals. - **AWS Control Tower:** Control Tower applies preventive (SCP/RCP), detective (AWS Config) and proactive (CloudFormation hook) controls to whole OUs of an AWS landing zone. - **Google Cloud IAM recommender:** The IAM recommender suggests removing or replacing IAM roles that principals have not used, and Policy Simulator shows the effect of a policy change before it is made. - **Turbot:** Guardrails sets a policy once at the root or a folder and corrects drift automatically from cloud change events across AWS, Azure, GCP, OCI and GitHub. - **Apono:** Apono grants just-in-time, task-scoped access to people, machines and AI agents in the native policy language of AWS, Azure, GCP, Kubernetes and databases, and revokes it automatically. - **Britive:** Britive creates task-scoped privileged access at runtime and removes it when the task ends, for humans, non-human identities and AI agents across AWS, Azure, GCP and OCI. - **AWS IAM Access Analyzer:** IAM Access Analyzer generates an IAM policy from up to 90 days of CloudTrail activity and recommends refined policies for unused permissions. - **Wiz:** Connects code, cloud, identities, network and runtime in one security graph to prioritize exploitable risk and route fixes to code owners. - **P0 Security:** Grants short-lived, scoped access on request through Slack, CLI or web with approvals, and applies the same runtime authorization to AI agents. - **CrowdStrike:** CrowdStrike ties cloud findings to adversary tradecraft it has observed to rank which exposures to fix first. - **Tenable Cloud Security:** Combines CIEM entitlement analysis and JIT access with posture management, IaC scanning and attack paths in one cloud platform. - **Entitle:** Entitle provides just-in-time cloud and SaaS access that auto-revokes, with approval workflows in Slack and Teams, access reviews and audit logs. - **Sysdig:** Sysdig uses runtime data to show which permissions are in use and generates access policies limited to that use. - **Prisma Cloud:** Calculates net effective permissions across clouds, including SCPs as input, and ties them to a broad platform with code scanning, network analysis and a large library of compliance standards. - **Microsoft:** Defender for Cloud evaluates effective permissions to identify which identities can reach sensitive or business-critical resources. ## Picks by situation - [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A company putting AI agents into production this year](/situations/ai-agents/): First: Act Security 5.6; then Sonrai Security 5.2; Act 1st - [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/): First: Sonrai Security 5.2 (joint with Act Security); Act joint 1st - [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/): First: Sonrai Security 5.5 (joint with Act Security); Act joint 1st ## Preventive guardrails A preventive guardrail is a policy that a cloud evaluates before a request takes effect and answers with a denial. That separates it from a detective control, which reports a violation after the resource exists. A service control policy that denies a leave-organization call, an Azure Policy definition with the deny effect and a Google Cloud organization policy constraint are all preventive. An alert on the same event is not. Each cloud evaluates its guardrails in a different place. AWS evaluates service control policies and resource control policies together with identity-based and resource-based policies, and the effective permissions are the intersection. Declarative policies work in the service control plane instead. Azure Policy checks create and update requests against assigned definitions. Google Cloud checks IAM deny policies before allow policies, and organization policy constraints, principal access boundaries and VPC Service Controls perimeters each add their own check. Because a guardrail answers with a denial, its cost is the legitimate work it stops. This library records, for each of 100 entries, what the guardrail blocks and what it breaks, with the source of both. The platforms section ranks sixteen products on how safely they take a guardrail from a draft to production: simulated first, shown against the real environment, and kept in place afterward. ## The Simulation Coverage Matrix Which platforms show the effect of a guardrail before it is applied, and which provider tools do the same for free: the [Simulation Coverage Matrix](/simulation/) sets the 16 platforms against the native dry-run and simulation methods of the three clouds. - [The Simulation Coverage Matrix](/simulation/) - [Every preventive mechanism, per cloud](/mechanisms/) - [All 100 guardrails](/library/) ## Guides and glossary - [AWS SCP examples explained: what each pattern blocks and breaks](/guides/aws-scp-examples-explained/) - [How SCP evaluation works: allow lists, deny lists, and the limits](/guides/how-scp-evaluation-works/) - [Resource control policies: what RCPs are and how they differ from SCPs](/guides/resource-control-policies/) - [Permissions boundaries: capping IAM users and roles, and delegation](/guides/permissions-boundaries/) - [Azure Policy as a guardrail: effects, enforcement mode, exemptions](/guides/azure-policy-as-a-guardrail/) - [Google Cloud organization policy: constraints, dry-run, IAM deny](/guides/google-cloud-organization-policy/) - [Glossary, 31 terms](/glossary/) - [30 questions answered](/faq/) ## Sources - AWS Organizations: service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - act.security: https://act.security (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/ --- # How platforms are ranked on putting a guardrail into production Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Sixteen platforms are scored 0 to 10 on seven criteria whose weights were fixed on 10 Nov 2026, before any score existed. Safe to enforce and one change for a class of risk carry 30 each. Totals are weighted means rounded to one decimal, and every fact is dated and linked to the page it came from. **What these weights produce.** Act is first at 5.9, ahead of Sonrai Security at 5.7. Scope: security governance, the guardrail layer. Cost governance is out. The seven criteria and their weights are published with this method and were fixed before any vendor was scored: safe to enforce 30, one change for a class of risk 30, removes access rather than only reporting it 15, stays fixed 10, uses the cloud's own controls 10, evidence for auditors 3, and identity, network and AI coverage together 2. The weights favor the question this site exists to answer: can the guardrail be shown to be safe before it ships, and does one policy cover a whole class of risk. A score is an integer from 0 to 10 for one criterion. The total is the sum of score times weight divided by the sum of the weights, rounded to one decimal. Every fact behind a score carries a quote, the page it came from and the date it was read, 10 Nov 2026. Every fact in the evidence files is labeled as coming from a vendor's own material or an analyst report about the vendor, so no score can exceed 7. A capability not found in the pages read is recorded as "none found on 2026-11-10", never as an absence. Each situation page ranks the same sixteen platforms on the mean of this site's weights and the situation's own weights, both fixed in advance. Act is scored by the same rules as every other platform. Where a platform's strength is documented, the page states it in one sentence from its evidence. ## The seven criteria and the weights this site uses | Criterion | What is scored | Weight | | --- | --- | --- | | Removes access, not only reports it | Does the product end with an enforced change that takes access away, or with a finding for someone else to act on | 15 | | Safe to enforce | Is each change simulated against the real environment before it ships, with the impact shown | 30 | | One change, a class of risk | Boundaries and guardrails that remove many paths at once, as opposed to permission-by-permission clean-up | 30 | | Stays fixed | Prevents drift: violations checked before production, exceptions handled in the team's own tools, with proof | 10 | | Identity, network and AI access together | Does the model of reachability cover all three, including AI workloads and agents | 2 | | Uses the cloud's own controls | Enforcement through native mechanisms the customer already owns, with nothing new in the request path | 10 | | Evidence for auditors | Mapped controls and continuous evidence for the named frameworks | 3 | Weights sum to 100, fixed on 10 Nov 2026 and committed before any vendor was scored. These weights produce the main ranking; situation pages use the mean described below. ## The scale and the rules, in brief Each criterion is scored 0 to 10 from the vendor's evidence: 0 means none found on the read date, 7 is the ceiling for anything resting on the vendor's own material, and 8 to 10 need a source the vendor does not control. Scoring was blind, from evidence with product names removed. The total is the weighted mean on the weights above, to one decimal; equal totals share a rank and are listed by the unrounded total. - Weights, criteria and the vendor set were committed before the first score, and the scores were produced from evidence packets with product names replaced by codes. - Equal totals at one decimal are shown as joint, and the order inside a tie follows the unrounded total. - The Simulation Coverage Matrix records what each platform shows before a guardrail ships, from the same evidence. ## Situation weights Each situation page ranks on the mean of this site's weights and the situation's own weights, both fixed here before scoring. | Situation | Removes | Safe | Class | Fixed | Id+Net+AI | Native | Audit | | --- | --- | --- | --- | --- | --- | --- | --- | | An AWS-heavy organization with hundreds of accounts and years of accumulated access | 20 | 25 | 27.5 | 12.5 | 3.5 | 7.5 | 4 | | A multi-cloud enterprise with separate teams per cloud | 17.5 | 22.5 | 22.5 | 15 | 8.5 | 12.5 | 1.5 | | A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53 | 17.5 | 22.5 | 20 | 15 | 3.5 | 7.5 | 14 | | A company putting AI agents into production this year | 17.5 | 22.5 | 20 | 10 | 18.5 | 7.5 | 4 | | A small cloud security team buried under posture findings it cannot clear | 22.5 | 22.5 | 27.5 | 12.5 | 3.5 | 7.5 | 4 | Published weights: the mean of this site's weights and the situation's own weights, both fixed before scoring. ## The ranking these weights produce | Rank | Platform | Total | Removes (15) | Safe (30) | Class (30) | Fixed (10) | Id+Net+AI (2) | Native (10) | Audit (3) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.9 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.7 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.2 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.1 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 5 | [Turbot](/platforms/turbot/) | 3.9 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 6 | [Apono](/platforms/apono/) | 3.6 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.0 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8= | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.9 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 8= | [Wiz](/platforms/wiz/) | 2.9 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 10= | [P0 Security](/platforms/p0/) | 2.3 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 10= | [CrowdStrike](/platforms/crowdstrike/) | 2.3 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 10= | [Tenable Cloud Security](/platforms/tenable/) | 2.3 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 13= | [Entitle](/platforms/entitle/) | 2.0 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 13= | [Sysdig](/platforms/sysdig/) | 2.0 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 1.7 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.1 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Weights in the header row sum to 100. Scores 0 to 10 from the evidence read 10 Nov 2026; weighted total to one decimal; "=" marks a joint rank, listed by the unrounded total. Act is first at 5.9, ahead of Sonrai Security at 5.7. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on safe to enforce rose by 1 point (from 5 to 6) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. ## Read dates and refresh Every vendor page behind this edition was read on 10 Nov 2026. Claims older than 60 days are re-read; a score changes only when a sourced fact changes, and every change is listed on the corrections page with the old and new values. ## Related pages - [30 questions about cloud governance guardrails](/faq/) - [The Guardrail Library: the data, downloadable](/datasets/) - [The Guardrail Library: 100 preventive guardrails for AWS, Azure and Google Cloud](/library/) ## Sources - Act platform page (read for the client facts, scored like every other vendor): https://act.security/platform (read 10 Nov 2026) - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/method/ --- # The 16 platforms, ranked by how safely they put a guardrail into production Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Sixteen platforms ranked on how safely they put a guardrail into production, from Act at 5.9 and Sonrai Security at 5.7 to Microsoft at 1.1. Each page shows the mechanisms it uses, how it tests a guardrail, the scores and the quoted evidence. | Rank | Platform | Total | Removes (15) | Safe (30) | Class (30) | Fixed (10) | Id+Net+AI (2) | Native (10) | Audit (3) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.9 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.7 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.2 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.1 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 5 | [Turbot](/platforms/turbot/) | 3.9 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 6 | [Apono](/platforms/apono/) | 3.6 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.0 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8= | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.9 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 8= | [Wiz](/platforms/wiz/) | 2.9 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 10= | [P0 Security](/platforms/p0/) | 2.3 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 10= | [CrowdStrike](/platforms/crowdstrike/) | 2.3 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 10= | [Tenable Cloud Security](/platforms/tenable/) | 2.3 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 13= | [Entitle](/platforms/entitle/) | 2.0 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 13= | [Sysdig](/platforms/sysdig/) | 2.0 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 1.7 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.1 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Weights in the header row sum to 100. Scores 0 to 10 from the evidence read 10 Nov 2026; weighted total to one decimal; "=" marks a joint rank, listed by the unrounded total. Act is first at 5.9, ahead of Sonrai Security at 5.7. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on safe to enforce rose by 1 point (from 5 to 6) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. ## Who should pick what - [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A company putting AI agents into production this year](/situations/ai-agents/): First: Act Security 5.6; then Sonrai Security 5.2; Act 1st - [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/): First: Sonrai Security 5.2 (joint with Act Security); Act joint 1st - [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/): First: Sonrai Security 5.5 (joint with Act Security); Act joint 1st ## Every platform page - [1. Act Security](/platforms/act/): Act removes broad classes of access by simulating guardrail and ring-fence policies against a digital twin and delivering them as pull requests to the customer's IaC using cloud-native controls. - [2. Sonrai Security](/platforms/sonrai/): Sonrai blocks unused privileged permissions, dormant identities, third-party access and unused services and regions through org-level cloud-native policies, and restores access through ChatOps approvals. - [3. AWS Control Tower](/platforms/aws-control-tower/): Control Tower applies preventive (SCP/RCP), detective (AWS Config) and proactive (CloudFormation hook) controls to whole OUs of an AWS landing zone. - [4. Google Cloud IAM recommender](/platforms/google-iam-recommender/): The IAM recommender suggests removing or replacing IAM roles that principals have not used, and Policy Simulator shows the effect of a policy change before it is made. - [5. Turbot](/platforms/turbot/): Guardrails sets a policy once at the root or a folder and corrects drift automatically from cloud change events across AWS, Azure, GCP, OCI and GitHub. - [6. Apono](/platforms/apono/): Apono grants just-in-time, task-scoped access to people, machines and AI agents in the native policy language of AWS, Azure, GCP, Kubernetes and databases, and revokes it automatically. - [7. Britive](/platforms/britive/): Britive creates task-scoped privileged access at runtime and removes it when the task ends, for humans, non-human identities and AI agents across AWS, Azure, GCP and OCI. - [8=. AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/): IAM Access Analyzer generates an IAM policy from up to 90 days of CloudTrail activity and recommends refined policies for unused permissions. - [8=. Wiz](/platforms/wiz/): Connects code, cloud, identities, network and runtime in one security graph to prioritize exploitable risk and route fixes to code owners. - [10=. P0 Security](/platforms/p0/): Grants short-lived, scoped access on request through Slack, CLI or web with approvals, and applies the same runtime authorization to AI agents. - [10=. CrowdStrike](/platforms/crowdstrike/): CrowdStrike ties cloud findings to adversary tradecraft it has observed to rank which exposures to fix first. - [10=. Tenable Cloud Security](/platforms/tenable/): Combines CIEM entitlement analysis and JIT access with posture management, IaC scanning and attack paths in one cloud platform. - [13=. Entitle](/platforms/entitle/): Entitle provides just-in-time cloud and SaaS access that auto-revokes, with approval workflows in Slack and Teams, access reviews and audit logs. - [13=. Sysdig](/platforms/sysdig/): Sysdig uses runtime data to show which permissions are in use and generates access policies limited to that use. - [15. Prisma Cloud](/platforms/prisma/): Calculates net effective permissions across clouds, including SCPs as input, and ties them to a broad platform with code scanning, network analysis and a large library of compliance standards. - [16. Microsoft](/platforms/microsoft/): Defender for Cloud evaluates effective permissions to identify which identities can reach sensitive or business-critical resources. ## Related pages - [Picks by situation](/situations/) - [Guides](/guides/) - [Glossary](/glossary/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/ --- # Act Security puts guardrails into production at 5.9, 1st of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Act Security is 1st of 16 on how safely a platform puts a guardrail into production, at 5.9 of 10. It scores 7 on safe to enforce and 7 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Act says it writes a guardrail as a policy in the cloud's own mechanism and delivers it as a pull request. Act's pages describe two shapes: broad explicit-deny guardrail campaigns for whole attack categories, with built-in exceptions, and explicit-allow ring-fencing around selected crown jewel applications and data. Act's blog places the controls at the AWS Organizations and service control policy level, and names Security Groups, NSGs, SCPs, RCPs and private endpoints as the native boundaries it refers to. Act says one policy change can remove thousands of access paths. Act says it tests before a change ships: each change is run against a digital twin built from cloud configuration data, then in audit mode against audit-log events, before enforcement. Act is read-only to the customer's cloud, so the change reaches production through the customer's Terraform or CloudFormation, a ticket or a coding agent. SACR's report on Act lists limits: rarely used or undocumented dependencies may go untested, and violations management and time-bound waivers were presented as near-term roadmap. Act's free IAM Simulate library evaluates a request against supplied identity, resource, SCP, RCP and permissions boundary policies, for AWS only. ## How Act puts a guardrail into production ### Safe to enforce: 7 of 10, weight 30 Each change is tested against a digital twin built from cloud configuration data and run in audit mode against audit-log events before enforcement. This is on the vendor's own material only, and rare dependencies may go untested. - "Simulation: Test changes against a digital twin to see their exact impact before enforcing, ensuring nothing breaks." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) - "Audit mode : Run changes in audit mode as long as needed to confirm no legitimate access is revoked." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) ### One change, a class of risk: 7 of 10, weight 30 Act runs broad explicit-deny guardrail campaigns for whole attack categories and explicit-allow ring-fencing, applied at the AWS Organizations and SCP level. This is on the vendor's own material only. - "Broad explicit-deny policies that block entire attack categories, including unauthorized cross-environment movement, data exfiltration, and external overexposure" (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) - "Least privilege, explicitly-allow policies, applied selectively to crown jewel applications and sensitive data." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) ### Uses the cloud's own controls: 6 of 10, weight 10 Act enforces through named native controls (security groups, NSGs, SCPs, RCPs, private endpoints) delivered via infrastructure as code, with no new infrastructure. For AI agents it also enforces on the agent and the gateway. - "Act enforces boundaries through your cloud's own native controls, using Infrastructure-as-Code. No agents and no new infrastructure to maintain." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) - "Act’s differentiation is therefore less about a unique enforcement mechanism, as it relies on the cloud’s native controls" (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) ### Stays fixed: 4 of 10, weight 10 Act documents drift detection with a diff against cloud versions, a one-line CI/CD blocking claim that does not say what is checked, and a Terraform provider for its own configuration. Exception handling in the customer's tools and rollback are credited only as roadmap or open discussion. Wiz and Sonrai also score 4. - "CI/CD integration: Block risk from shipping in CI/CD, with no downtime or disruption to operations." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) - "let users request blocked access directly within their existing tools, so security can approve or deny in one click." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) ### Removes access, not only reports it: 3 of 10, weight 15 Changes are delivered as infrastructure-as-code pull requests that the customer applies. The product is read-only, and direct enforcement is still under discussion. - "Simulate each policy change first to validate its impact before it reaches production, then apply it through your own cloud-native controls." (Act's own material, [read 10 Nov 2026](https://act.security/?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) - "Native enforcement : Deploy directly as pull requests to your existing Infrastructure-as-Code (Terraform, CloudFormation) using the cloud-native controls you already own." (Act's own material, [read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 7 | 2.10 | | One change, a class of risk | 30 | 7 | 2.10 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 6 | 0.12 | | Uses the cloud's own controls | 10 | 6 | 0.60 | | Evidence for auditors | 3 | 3 | 0.09 | Weighted total 5.9 (5.86 before rounding) on this site's weights. ## Act in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 1st | 5.6 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 1st | 5.6 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 1st | 5.6 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 1st | 5.2 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 1st | 5.5 | Sonrai Security 5.5 | ## Strongest, in one sentence Act removes broad classes of access by simulating guardrail and ring-fence policies against a digital twin and delivering them as pull requests to the customer's IaC using cloud-native controls. ([source](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=platform)) ## Related pages - [Sonrai Security for guardrails: 2nd of 16 at 5.7](/platforms/sonrai/) - [AWS Control Tower for guardrails: 3rd of 16 at 4.2](/platforms/aws-control-tower/) - [Google Cloud IAM recommender for guardrails: 4th of 16 at 4.1](/platforms/google-iam-recommender/) ## Sources - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) - act.security/resources/ai-agent-security-access-chain: https://act.security/resources/ai-agent-security-access-chain (read 10 Nov 2026) - act.security/resources/zero-trust-ai-agents: https://act.security/resources/zero-trust-ai-agents (read 10 Nov 2026) - raw.githubusercontent.com/act-security/terraform-provider-act/main/README.md: https://raw.githubusercontent.com/act-security/terraform-provider-act/main/README.md (read 10 Nov 2026) - act.security/resources/network-and-identity-each-see-half-your-cloud-your-attack: https://act.security/resources/network-and-identity-each-see-half-your-cloud-your-attacker-sees-both (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/act/ --- # Sonrai Security puts guardrails into production at 5.7, 2nd of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Sonrai Security is 2nd of 16 on how safely a platform puts a guardrail into production, at 5.7 of 10. It scores 5 on safe to enforce and 7 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Sonrai's Cloud Permissions Firewall builds a guardrail from observed usage. Sonrai says its baseline comes from permission usage across humans, machines, roles and AI agents, that actively used permissions are exempted, and that the exemption list is shown before deployment. The guardrails are organization-level policies in the cloud's own mechanisms: Sonrai says it enforces with SCPs, RCPs, Azure RBAC and Google Cloud deny policies, with no proxy or broker. One action can apply a global default deny, and unused services and regions are blocked with centralized policies. Sonrai is onboarded at the organization level with a read-only role, and the customer chooses where controls are deployed. Sonrai's pages say nothing enforces until the customer deploys the policies, and that blocked access is reversible and logged. The pages do not describe a simulation of the change itself. Third-party access control is described only with AWS resource control policies, and the Slack request workflow starts after a restriction is in place. Sonrai's documented strength is that it blocks unused privileges through organization-level policies in the cloud's own mechanisms. ## How Sonrai puts a guardrail into production ### Safe to enforce: 5 of 10, weight 30 Policies are built from observed usage, in-use permissions are exempted, and the exemption list is shown before deployment. The change is reversible, but it is not described as a simulation of the change. - "Policies are built from your usage data. Anything in use is exempted before it’s written. You see the exemption list first." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com)) - "All reversible and logged." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com)) ### One change, a class of risk: 7 of 10, weight 30 Sonrai offers a global default deny in one action, AWS resource control policy (RCP) third-party boundaries at any level, service and region blocks, and organization-level just-in-time access across AWS, GCP and Azure. The RCP feature is AWS only, on the vendor's own material only. - "Global default deny is applied with one click, while actively used permissions remain available." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com/use-cases/least-privilege)) - "Automatically block future third-party access through default deny policies" (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com/platform)) ### Uses the cloud's own controls: 7 of 10, weight 10 Enforcement uses SCPs, RCPs, Azure RBAC and GCP deny policies, with no proxy, broker or connector. This rests on the vendor's own material only. - Enforcement uses SCPs, RCPs, Azure RBAC and GCP deny policies, and the vendor states there is no proxy or broker. (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com)) - Limit: "When the allocated time is over, access is automatically revoked." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com)) ### Stays fixed: 4 of 10, weight 10 Blocked attempts open Slack requests, and changes are recorded in ITSM. There is no pre-production check, and the workflow starts only after the restriction is applied. - "Blocked usage attempts open a request, routes it in Slack, and updates the policy for you." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com)) - "Request and approval processes live in the chat tools you already use today and all access changes are recorded in your ITSM." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com/use-cases/least-privilege)) ### Removes access, not only reports it: 5 of 10, weight 15 Once the customer deploys it, organization-level policies block unused privileges and quarantine dormant identities automatically. Removal of unused permissions is only suggested. - "The Cloud Permissions Firewall continuously analyzes real permission usage across all identities and automatically blocks unused privileges using cloud-native, org-level policies." (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com/platform)) - "Unused permissions are restricted, dormant identities are quarantined, and risky services or regions are blocked" (Sonrai's own material, [read 10 Nov 2026](https://sonraisecurity.com/platform)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 5 | 1.50 | | One change, a class of risk | 30 | 7 | 2.10 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 7 | 0.70 | | Evidence for auditors | 3 | 4 | 0.12 | Weighted total 5.7 (5.65 before rounding) on this site's weights. ## Sonrai in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 2nd | 5.5 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 2nd | 5.5 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 2nd | 5.2 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 1st | 5.2 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 1st | 5.5 | Sonrai Security 5.5 | ## Strongest, in one sentence Sonrai blocks unused privileged permissions, dormant identities, third-party access and unused services and regions through org-level cloud-native policies, and restores access through ChatOps approvals. ([source](https://sonraisecurity.com/platform)) ## Related pages - [AWS Control Tower for guardrails: 3rd of 16 at 4.2](/platforms/aws-control-tower/) - [Google Cloud IAM recommender for guardrails: 4th of 16 at 4.1](/platforms/google-iam-recommender/) - [Turbot for guardrails: 5th of 16 at 3.9](/platforms/turbot/) ## Sources - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) - sonraisecurity.com/use-cases/ai-agent-security: https://sonraisecurity.com/use-cases/ai-agent-security (read 10 Nov 2026) - sonraisecurity.com/use-cases/just-in-time-access: https://sonraisecurity.com/use-cases/just-in-time-access (read 10 Nov 2026) - sonraisecurity.com/pricing: https://sonraisecurity.com/pricing (read 10 Nov 2026) - sonraisecurity.com/use-cases/third-party-access: https://sonraisecurity.com/use-cases/third-party-access/ (read 10 Nov 2026) - sonraisecurity.com/firewall-slack-integration: https://sonraisecurity.com/firewall-slack-integration/ (read 10 Nov 2026) - sonraisecurity.com/comparisons/wiz: https://sonraisecurity.com/comparisons/wiz/ (read 10 Nov 2026) - sonraisecurity.com/resource/iso27002: https://sonraisecurity.com/resource/iso27002/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/sonrai/ --- # AWS Control Tower puts guardrails into production at 4.2, 3rd of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS Control Tower is 3rd of 16 on how safely a platform puts a guardrail into production, at 4.2 of 10. It scores 1 on safe to enforce and 6 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. AWS Control Tower puts guardrails on whole OUs from a catalog. AWS says Control Catalog lists more than 750 managed controls, and preventive controls are implemented with SCPs and RCPs, detective controls with AWS Config rules and proactive controls with CloudFormation hooks. Once a preventive control is enabled on an OU it is enforced on every member account in it. The root user and administrators in the management account can do work that controls would otherwise deny, and a limit of 10 SCPs per OU applies in a Control Tower landing zone. On testing, the Control Tower documentation recommends testing proactive controls individually before enabling them, and no simulation of impact is described. Control Tower scans managed SCPs, RCPs and declarative policies daily for drift, and sends a notification only for the first instance of drift on a resource. Some controls take parameters such as ExemptedPrincipalArns, so exemptions live in the template. SCP and Config controls must be removed by hand when an account moves between OUs. Control Tower works in AWS only. ## How Control Tower puts a guardrail into production ### Safe to enforce: 1 of 10, weight 30 Users are told to test proactive controls themselves, and detective controls report after the fact. - Limit: "We recommend that you test proactive controls individually before you enable them in your environment." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html)) - Limit: "Detective controls detect specific events when they occur" (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) ### One change, a class of risk: 6 of 10, weight 30 Controls apply to whole OUs through SCPs and RCPs, from a catalog of more than 750 controls. Limits include a management-account bypass, a 10 SCP per OU limit, account limits, and AWS only. - "A control applies to an entire organizational unit (OU), and every AWS account within the OU is affected by the control." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/controls.html)) - "In AWS Control Tower preventive controls are implemented with service control policies (SCPs) and resource control policies (RCPs)." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) ### Uses the cloud's own controls: 6 of 10, weight 10 Enforcement uses SCPs and RCPs, AWS Config and CloudFormation hooks, with nothing in the workload request path. It is AWS only, with a management-account bypass. - "Detective controls are implemented with AWS Config rules. Proactive controls are implemented with CloudFormation hooks." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) - "AWS Config tracks activity on your AWS account resources in target organizational units that you specify and powers detective controls." (Control Tower's own material, [read 10 Nov 2026](https://aws.amazon.com/controltower/features/)) ### Stays fixed: 5 of 10, weight 10 A CloudFormation hook checks before provisioning, SCP and RCP drift is scanned daily, and exemptions live in templates. Hooks cover CloudFormation only, and Account Factory for Terraform has manual and regional limits. - "Proactive controls check whether resources are compliant with your company policies and objectives, before the resources are provisioned in your accounts." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) - Limit: "these controls may not affect requests that are made directly to services through the AWS console, through AWS APIs, or through other means such as AWS SDKs, or other Infrastructure-as-Code (IaC) tools" (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html)) ### Removes access, not only reports it: 5 of 10, weight 15 Once enabled on an OU, preventive controls deny actions on member accounts. Drift is only notified, SCP and Config controls need manual removal when accounts move between OUs, and this is AWS only. - "Preventive controls prevent actions from occurring." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) - "A preventive control ensures that your accounts maintain compliance, because it disallows actions that lead to policy violations." (Control Tower's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 1 | 0.30 | | One change, a class of risk | 30 | 6 | 1.80 | | Stays fixed | 10 | 5 | 0.50 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 6 | 0.60 | | Evidence for auditors | 3 | 4 | 0.12 | Weighted total 4.2 (4.15 before rounding) on this site's weights. ## Control Tower in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 3rd | 4.3 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 3rd | 4.4 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 3rd | 4.2 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 4th | 4.2 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 3rd | 4.4 | Sonrai Security 5.5 | ## Strongest, in one sentence Control Tower applies preventive (SCP/RCP), detective (AWS Config) and proactive (CloudFormation hook) controls to whole OUs of an AWS landing zone. ([source](https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html)) ## Related pages - [Google Cloud IAM recommender for guardrails: 4th of 16 at 4.1](/platforms/google-iam-recommender/) - [Turbot for guardrails: 5th of 16 at 3.9](/platforms/turbot/) - [Apono for guardrails: 6th of 16 at 3.6](/platforms/apono/) ## Sources - docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html: https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.htm: https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/drift.html: https://docs.aws.amazon.com/controltower/latest/userguide/drift.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html: https://docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/controls.html: https://docs.aws.amazon.com/controltower/latest/userguide/controls.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/what-is-control-tower.html: https://docs.aws.amazon.com/controltower/latest/userguide/what-is-control-tower.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/region-how.html: https://docs.aws.amazon.com/controltower/latest/userguide/region-how.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/controlreference/security-hub-controls.h: https://docs.aws.amazon.com/controltower/latest/controlreference/security-hub-controls.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/limits.html: https://docs.aws.amazon.com/controltower/latest/userguide/limits.html (read 10 Nov 2026) - aws.amazon.com/controltower/features: https://aws.amazon.com/controltower/features/ (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html: https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/enable-controls.html: https://docs.aws.amazon.com/controltower/latest/userguide/enable-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/aws-control-tower/ --- # Google Cloud IAM recommender puts guardrails into production at 4.1, 4th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud IAM recommender is 4th of 16 on how safely a platform puts a guardrail into production, at 4.1 of 10. It scores 6 on safe to enforce and 3 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. The Google Cloud IAM recommender and Policy Intelligence give the most native testing for Google Cloud guardrails. Policy Simulator for Organization Policy lists the resources that would violate a proposed policy before it is enforced, and Policy Simulator for allow policies shows how a change affects principals' access. Simulators also exist for deny and principal access boundary policies, and the Organization Policy recommender analyzes existing configurations and recommends which constraints to enforce. Google says each simulator evaluates only one policy type. The recommender itself works role by role: a role recommendation addresses one IAM role granted to principals, and a person reviews each one to apply or dismiss it. Google tells users to have a reviewer assess impact before applying recommendations. A pre-deploy Terraform check, gcloud beta terraform vet, validates a plan against an organization policy library, but it is a separate Preview tool. The IAM recommender works in Google Cloud only, and AI workload coverage: none found on 2026-11-10. ## How IAM recommender puts a guardrail into production ### Safe to enforce: 6 of 10, weight 30 Simulators show how an allow-policy change affects principals' access, and list resources that would violate a proposed organization policy before enforcement. Each simulator covers one policy type, and only Google Cloud. - "Policy Simulator for Identity and Access Management allow policies lets you see how a change to an allow policy might affect a principal's access before you commit to making the change." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/iam-simulator-overview)) - "Policy Simulator provides a list of resources that violate the proposed policy before it is enforced, allowing you to reconfigure those resources, request exceptions, or change the scope of your organization policy" (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/overview)) ### One change, a class of risk: 3 of 10, weight 30 Simulators exist for deny and Principal Access Boundary policies, and organization policies are recommended, but clean-up is done role by role. - "Policy Simulator for deny policies Policy Simulator for Principal Access Boundary policies" (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/overview)) - "It analyzes existing organization policy configurations and generates recommendations for which organization policies to enforce." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/organization-policy-recommendations-overview)) ### Uses the cloud's own controls: 5 of 10, weight 10 A recommendation is an operation that removes a member from native IAM policy bindings. It works in one cloud only. - "action : remove , path : /iamPolicy/bindings//members/" (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/role-recommendations-overview)) ### Stays fixed: 2 of 10, weight 10 Recommendation states are tracked inside the product. The pre-deploy Terraform check is a separate tool, and no ticketing or chat integration was found. - "gcloud beta terraform vet lets you validate your terraform plan JSON against your organization's POLICY_LIBRARY_REPO ." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/docs/terraform/policy-validation/validate-policies)) - Limit: "You can mark a recommendation as dismissed to indicate that you don't intend to apply the recommended changes to the associated resource or that you don't want to continue to see the recommendation." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/recommender/docs/using-api)) ### Removes access, not only reports it: 4 of 10, weight 15 A person checks and applies role recommendations in the product. Organization policy recommendations are suggestions only. - Limit: "The IAM recommender does not apply recommendations automatically. Instead, you must review your recommendations and decide whether to apply or dismiss them." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/role-recommendations-overview)) - "You can apply recommendations to replace or remove roles, or dismiss recommendations that you don't want to apply." (IAM recommender's own material, [read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/role-recommendations-overview)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 4 | 0.60 | | Safe to enforce | 30 | 6 | 1.80 | | One change, a class of risk | 30 | 3 | 0.90 | | Stays fixed | 10 | 2 | 0.20 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 5 | 0.50 | | Evidence for auditors | 3 | 2 | 0.06 | Weighted total 4.1 (4.14 before rounding) on this site's weights. ## IAM recommender in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 5th | 4.0 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 5th | 4.0 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 5th | 4.0 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 5th | 3.7 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 5th | 3.9 | Sonrai Security 5.5 | ## Strongest, in one sentence The IAM recommender suggests removing or replacing IAM roles that principals have not used, and Policy Simulator shows the effect of a policy change before it is made. ([source](https://cloud.google.com/policy-intelligence/docs/role-recommendations-overview)) ## Related pages - [Turbot for guardrails: 5th of 16 at 3.9](/platforms/turbot/) - [Apono for guardrails: 6th of 16 at 3.6](/platforms/apono/) - [Britive for guardrails: 7th of 16 at 3.0](/platforms/britive/) ## Sources - cloud.google.com/policy-intelligence/docs/role-recommendations-overview: https://cloud.google.com/policy-intelligence/docs/role-recommendations-overview (read 10 Nov 2026) - cloud.google.com/policy-intelligence/docs/organization-policy-recommendations-ov: https://cloud.google.com/policy-intelligence/docs/organization-policy-recommendations-overview (read 10 Nov 2026) - cloud.google.com/policy-intelligence/docs/iam-simulator-overview: https://cloud.google.com/policy-intelligence/docs/iam-simulator-overview (read 10 Nov 2026) - cloud.google.com/policy-intelligence/docs/overview: https://cloud.google.com/policy-intelligence/docs/overview (read 10 Nov 2026) - cloud.google.com/docs/terraform/policy-validation/validate-policies: https://cloud.google.com/docs/terraform/policy-validation/validate-policies (read 10 Nov 2026) - cloud.google.com/recommender/docs/using-api: https://cloud.google.com/recommender/docs/using-api (read 10 Nov 2026) - cloud.google.com/recommender/docs/bq-export/export-recommendations-to-bq: https://cloud.google.com/recommender/docs/bq-export/export-recommendations-to-bq (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/google-iam-recommender/ --- # Turbot puts guardrails into production at 3.9, 5th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Turbot is 5th of 16 on how safely a platform puts a guardrail into production, at 3.9 of 10. It scores 2 on safe to enforce and 5 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Turbot Guardrails sets a policy once at the root or a folder and every account and resource below inherits it, including new ones. Turbot's Trusted Access guardrail lets the user define whom and what is trusted and enforce those limits. Enforcement is done by Turbot itself, through a role it holds in each target account that can describe and modify resources, not through a cloud policy object it creates. Turbot's FAQ places AWS SCPs alongside Guardrails as preventative controls, so the two work together. Turbot's testing path is its modes. A control can be set to monitor, which shows an Alarm state for administrators to review, and Turbot's policy pack example keeps check-mode policies in one pack and enforcement policies in another. In Enforce mode Turbot applies the fix itself, from a stream of change events. A simulation of an enforcement's effect before it runs: none found on 2026-11-10. Policy exceptions stay visible and can carry an expiration date, and policy can be managed in Terraform. ## How Turbot puts a guardrail into production ### Safe to enforce: 2 of 10, weight 30 Monitor and check-mode policies let administrators look over alarms before enforcing. No simulation of an enforcement's impact is described. - "have a policy/ collection of policies configured such that Guardrails is only monitoring cloud resources rather than remediating. Controls in alarm should be investigated by administrators, as enforcement requires review." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/docs/concepts/controls)) - "The check-mod policies in Enterprise Checks are set lower because we are not yet ready for enforcements." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/docs/guides/policy-packs)) ### One change, a class of risk: 5 of 10, weight 30 Policies set on a folder or root are inherited by all accounts and new resources, and Trusted Access guardrails define trust limits. Enforcement is done by the product itself, not by a named cloud deny boundary. - "Policies set on a folder or the root flow down to every account and resource below it, including new ones." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/pricing)) - "Trusted Access allow you to define whom and what you trust and enforce those limitations on your cloud resources." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/docs/concepts/guardrails)) ### Uses the cloud's own controls: 3 of 10, weight 10 The product's own role modifies resources; enforcement does not use cloud policy objects. Service control policies (SCPs) are positioned alongside Turbot, not used by it. - Limit: Enforcement is done by Guardrails itself, using a role it holds in each target account that must be able to describe and modify resources, rather than by a cloud policy object it creates. (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/docs/concepts/controls)) - "SCPs are preventative controls, Guardrails can also provide detective and corrective controls across unapproved regions" (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/docs/faq/guardrails-and-aws-scps)) ### Stays fixed: 5 of 10, weight 10 Drift is checked within seconds, exceptions are visible with an expiry, and policy can be managed in Terraform. There is no pre-production check, and exceptions live in the product rather than in the team's own tools. - "Override a required policy on a specific account or resource when it has a valid reason to be different. The exception stays visible and auditable." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/pricing)) - "Give any policy setting or exception an expiration date. When it expires, the resource goes back to the inherited setting." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/pricing)) ### Removes access, not only reports it: 5 of 10, weight 15 In Enforce mode Turbot fixes misconfigurations automatically from change events, such as blocking public access. The examples are configuration fixes rather than access removal, and no per-cloud enforcement detail is given. - "Every resource change streams into the event loop and misconfigurations are fixed in seconds , not routed to a ticket queue." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails)) - "In Enforce mode, Turbot Guardrails fixes misconfigurations automatically, for example by enabling versioning, applying tags, or blocking public access." (Turbot's own material, [read 10 Nov 2026](https://turbot.com/guardrails/pricing)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 2 | 0.60 | | One change, a class of risk | 30 | 5 | 1.50 | | Stays fixed | 10 | 5 | 0.50 | | Identity, network and AI access together | 2 | 5 | 0.10 | | Uses the cloud's own controls | 10 | 3 | 0.30 | | Evidence for auditors | 3 | 6 | 0.18 | Weighted total 3.9 (3.93 before rounding) on this site's weights. ## Turbot in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 4th | 4.1 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 4th | 4.1 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 3rd | 4.2 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 3rd | 4.3 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 4th | 4.2 | Sonrai Security 5.5 | ## Strongest, in one sentence Guardrails sets a policy once at the root or a folder and corrects drift automatically from cloud change events across AWS, Azure, GCP, OCI and GitHub. ([source](https://turbot.com/guardrails)) ## Related pages - [Apono for guardrails: 6th of 16 at 3.6](/platforms/apono/) - [Britive for guardrails: 7th of 16 at 3.0](/platforms/britive/) - [AWS IAM Access Analyzer for guardrails: joint 8th of 16 at 2.9](/platforms/aws-iam-access-analyzer/) ## Sources - turbot.com/guardrails: https://turbot.com/guardrails (read 10 Nov 2026) - turbot.com/guardrails/pricing: https://turbot.com/guardrails/pricing (read 10 Nov 2026) - turbot.com/guardrails/docs/concepts/guardrails: https://turbot.com/guardrails/docs/concepts/guardrails (read 10 Nov 2026) - turbot.com/security: https://turbot.com/security (read 10 Nov 2026) - turbot.com/guardrails/docs/concepts/controls: https://turbot.com/guardrails/docs/concepts/controls (read 10 Nov 2026) - turbot.com/guardrails/docs/guides/policy-packs: https://turbot.com/guardrails/docs/guides/policy-packs (read 10 Nov 2026) - turbot.com/guardrails/docs/faq/guardrails-and-aws-scps: https://turbot.com/guardrails/docs/faq/guardrails-and-aws-scps (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/turbot/ --- # Apono puts guardrails into production at 3.6, 6th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Apono is 6th of 16 on how safely a platform puts a guardrail into production, at 3.6 of 10. It scores 3 on safe to enforce and 3 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Apono's closest control to a guardrail works on one principal at a time. Apono says Access Discovery can quarantine a dormant principal by placing it in an AWS permission boundary through an automatic access flow, kept until an admin disables the flow or deletes the principal. For an overprivileged policy, Apono gives a deny rule as JSON that, per Apono's documentation, the user creates in AWS. Right Sizing works per access flow, so Apono removes unused permissions one flow at a time rather than with one policy over a class of access. Apono's main product grants just-in-time access in the native policy language of AWS, Azure, GCP, Kubernetes and databases, and revokes it automatically. For changes to access flows made through the Apono Admin MCP server, Apono shows a preview of who gets what access, for how long and who approves, and saves nothing until confirmed. The quarantine action has no described simulation, and delete and right-size fixes are code the user runs. Apono Access Flows can be defined in Terraform. ## How Apono puts a guardrail into production ### Safe to enforce: 3 of 10, weight 30 A preview of who gets what access is shown before access-flow changes are saved, which is a narrow slice. Remediation actions have no simulation, and some fixes are code the user runs. - "Previews the proposed change. Apono applies only changes that have been previewed. The preview summarizes who receives what access, for how long, and who approves it. It also warns if the change expands access." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md)) - Limit: "On the Quarantine Principal tab, click Remediate to limit access within the dedicated access flow." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/getting-started/access-discovery/investigate-and-resolve-overprivileged-access.md)) ### One change, a class of risk: 3 of 10, weight 30 Apono can quarantine a single principal with an AWS permission boundary. Deny policies for overprivileged policies are created by the user, and right-sizing is done per flow. - "This approach uses an Automatic Access Flow to restrict a principal's access using an AWS Permission Boundary until it can be reviewed or safely deleted." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/getting-started/access-discovery/investigate-and-resolve-overprivileged-access.md)) - Limit: "In your AWS environment, create a deny rule using the Apono-provided JSON. This rule will prevent the principal from using the unused actions detected in its policy." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/getting-started/access-discovery/investigate-and-resolve-overprivileged-access.md)) ### Uses the cloud's own controls: 5 of 10, weight 10 Permissions are written in the native policy languages of AWS, Azure and GCP. A connector mediates grants, and whether a gateway sits in the request path is not established. - "Apono creates permissions at request time in the native policy language of AWS, Azure, GCP, Kubernetes, and your databases." (Apono's own material, [read 10 Nov 2026](https://www.apono.io)) - Limit: "Connectors are the components that mediate between Apono and your resources to sync data from cloud applications and grant and revoke access permissions." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/getting-started)) ### Stays fixed: 3 of 10, weight 10 Requests and approvals run in Slack, Teams, the CLI and Terraform-defined flows. No pre-production check or drift handling is described. - "Engineers request and receive access through Slack, the CLI, service workflows, or Apono Assist, wherever they already work." (Apono's own material, [read 10 Nov 2026](https://www.apono.io)) - "Create Access Flows with our declarative UI or in Terraform, and your developers can use Slack, Teams or CLI to request and approve access." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/why-choose-apono)) ### Removes access, not only reports it: 5 of 10, weight 15 Just-in-time access is created at request time and revoked automatically. Fixes for standing access (delete or right-size) are code that the user runs. - "Apono's permissions are just-in-time and also ephemeral. Access is automatically revoked when no longer needed." (Apono's own material, [read 10 Nov 2026](https://docs.apono.io/docs/why-choose-apono)) - "Apono creates IAM roles, permissions, and access policies on the fly at request time, scoped to the exact need and in the native policy language of your cloud platform." (Apono's own material, [read 10 Nov 2026](https://www.apono.io)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 3 | 0.90 | | One change, a class of risk | 30 | 3 | 0.90 | | Stays fixed | 10 | 3 | 0.30 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 5 | 0.50 | | Evidence for auditors | 3 | 4 | 0.12 | Weighted total 3.6 (3.55 before rounding) on this site's weights. ## Apono in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 6th | 3.6 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 6th | 3.7 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 6th | 3.7 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 5th | 3.7 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 6th | 3.7 | Sonrai Security 5.5 | ## Strongest, in one sentence Apono grants just-in-time, task-scoped access to people, machines and AI agents in the native policy language of AWS, Azure, GCP, Kubernetes and databases, and revokes it automatically. ([source](https://www.apono.io)) ## Related pages - [Britive for guardrails: 7th of 16 at 3.0](/platforms/britive/) - [AWS IAM Access Analyzer for guardrails: joint 8th of 16 at 2.9](/platforms/aws-iam-access-analyzer/) - [Wiz for guardrails: joint 8th of 16 at 2.9](/platforms/wiz/) ## Sources - docs.apono.io/docs/why-choose-apono: https://docs.apono.io/docs/why-choose-apono (read 10 Nov 2026) - apono.io: https://www.apono.io (read 10 Nov 2026) - docs.apono.io/docs/getting-started: https://docs.apono.io/docs/getting-started (read 10 Nov 2026) - docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md: https://docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md (read 10 Nov 2026) - docs.apono.io/docs/getting-started/access-discovery/investigate-and-resolve-over: https://docs.apono.io/docs/getting-started/access-discovery/investigate-and-resolve-overprivileged-access.md (read 10 Nov 2026) - docs.apono.io/docs/access-flows/manage-access-flows/right-sizing.md: https://docs.apono.io/docs/access-flows/manage-access-flows/right-sizing.md (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/apono/ --- # Britive puts guardrails into production at 3.0, 7th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Britive is 7th of 16 on how safely a platform puts a guardrail into production, at 3.0 of 10. It scores 1 on safe to enforce and 3 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Britive's pages describe scoping access, not writing a deny. A Britive profile for AWS can be associated with an organization unit, an account, or groups of either, so one profile definition covers many accounts. That scopes just-in-time access that Britive grants; Britive's pages describe no organization-wide boundary, and a profile holds one role. Britive creates privilege at runtime and removes it when the task ends, across AWS, Azure, GCP and OCI. Britive's testing evidence is a Validate button on inline AWS policies, and a check of profile policy data when a profile is cloned. Britive's page does not say what validation covers or that impact is shown. Permissions provisioned from a template show as out of sync if changed in AWS, and the difference is resolved at the next checkout. Britive says the decision is enforced natively at the resource, but its MCP Gateway sits in the path of agent tool calls. ## How Britive puts a guardrail into production ### Safe to enforce: 1 of 10, weight 30 Only a policy Validate button and profile-data validation exist. No impact simulation is described. - "Enter the Name and the Policy code in JSON format in the Create Policy page. Click Validate to validate the policy details." (Britive's own material, [read 10 Nov 2026](https://docs.britive.com/docs/profilemgmt-managing-profiles.md)) - "Britive validates the profile policy data and prompts the user for confirmation before skipping the erroneous policies." (Britive's own material, [read 10 Nov 2026](https://docs.britive.com/docs/profilemgmt-managing-profiles.md)) ### One change, a class of risk: 3 of 10, weight 30 A profile can be associated with OUs or accounts, which scopes just-in-time grants, but there is no deny or boundary and each profile holds one role. - "One way is to select an organization unit or an account, or a group of organization units or accounts, under Associations in Profiles." (Britive's own material, [read 10 Nov 2026](https://docs.britive.com/docs/britive-profile-management-for-aws.md)) - Limit: "for AWS, the user can add only one role per profile" (Britive's own material, [read 10 Nov 2026](https://docs.britive.com/docs/profilemgmt-managing-profiles.md)) ### Uses the cloud's own controls: 5 of 10, weight 10 The decision is enforced natively at the resource, agentless and proxyless, but the MCP gateway sits in the path of agent calls. - "The decision lands natively at the resource itself." (Britive's own material, [read 10 Nov 2026](https://www.britive.com)) - "Britive's agentless, proxyless architecture breaks that model." (Britive's own material, [read 10 Nov 2026](https://www.britive.com/company/britive-advantages)) ### Stays fixed: 3 of 10, weight 10 Requests run in Slack and Teams, with generic CI/CD and ITSM integrations. No pre-production policy check or drift handling is described. - "Request, approve, and check in privileged access from Slack and Microsoft Teams, under the same Britive policies, approvals, and expiry as the console." (Britive's own material, [read 10 Nov 2026](https://www.britive.com/use-cases)) - The platform integrates with CI/CD pipelines and automation tools. (Britive's own material, [read 10 Nov 2026](https://www.britive.com/platform)) ### Removes access, not only reports it: 5 of 10, weight 15 Just-in-time privilege is created and removed automatically when the task ends, named for AWS, Azure, GCP and OCI. - "Privileged access is created at the moment it's needed and removed when the task ends, automatically, across AWS, Azure, GCP, and OCI." (Britive's own material, [read 10 Nov 2026](https://www.britive.com)) - "Granular least-privilege provisioned at runtime, removed automatically." (Britive's own material, [read 10 Nov 2026](https://www.britive.com)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 1 | 0.30 | | One change, a class of risk | 30 | 3 | 0.90 | | Stays fixed | 10 | 3 | 0.30 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 5 | 0.50 | | Evidence for auditors | 3 | 4 | 0.12 | Weighted total 3.0 (2.95 before rounding) on this site's weights. ## Britive in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 7th | 3.1 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 7th | 3.3 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 8th | 3.3 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 8th | 3.2 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 7th | 3.2 | Sonrai Security 5.5 | ## Strongest, in one sentence Britive creates task-scoped privileged access at runtime and removes it when the task ends, for humans, non-human identities and AI agents across AWS, Azure, GCP and OCI. ([source](https://www.britive.com)) ## Related pages - [AWS IAM Access Analyzer for guardrails: joint 8th of 16 at 2.9](/platforms/aws-iam-access-analyzer/) - [Wiz for guardrails: joint 8th of 16 at 2.9](/platforms/wiz/) - [P0 Security for guardrails: joint 10th of 16 at 2.3](/platforms/p0/) ## Sources - britive.com: https://www.britive.com (read 10 Nov 2026) - britive.com/use-cases: https://www.britive.com/use-cases (read 10 Nov 2026) - britive.com/platform: https://www.britive.com/platform (read 10 Nov 2026) - docs.britive.com: https://docs.britive.com (read 10 Nov 2026) - britive.com/company/britive-advantages: https://www.britive.com/company/britive-advantages (read 10 Nov 2026) - docs.britive.com/docs/profilemgmt-managing-profiles.md: https://docs.britive.com/docs/profilemgmt-managing-profiles.md (read 10 Nov 2026) - docs.britive.com/docs/managed-permissions.md: https://docs.britive.com/docs/managed-permissions.md (read 10 Nov 2026) - docs.britive.com/docs/britive-profile-management-for-aws.md: https://docs.britive.com/docs/britive-profile-management-for-aws.md (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/britive/ --- # AWS IAM Access Analyzer puts guardrails into production at 2.9, joint 8th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS IAM Access Analyzer is joint 8th of 16 on how safely a platform puts a guardrail into production, at 2.9 of 10. It scores 4 on safe to enforce and 1 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. AWS IAM Access Analyzer generates and checks policies; applying a guardrail is not among the capabilities AWS lists. It builds an IAM policy from up to 90 days of CloudTrail activity and recommends refined policies for unused permissions, which the user attaches. The generated policy omits action-level data events and leaves out iam:PassRole. Access Analyzer reads resource control policies in its analysis, picking up an RCP change at the next periodic scan within 24 hours, but AWS lists no organization-wide boundary generation. Access Analyzer carries the most direct pre-deployment checks in the AWS set. AWS says findings can preview a policy's effect on public and cross-account access before resource permissions are deployed, and custom policy checks compare an updated policy against the existing version for new access. A customer quote on the AWS page says the checks in CI/CD trigger approval workflows. The service analyzes logs and policies and is not described as sitting in the request path. Access Analyzer covers AWS only. ## How Access Analyzer puts a guardrail into production ### Safe to enforce: 4 of 10, weight 30 The tool previews a policy's effect on public and cross-account access and checks new access against the existing version. It does not simulate the impact of removal, and generated policies omit data events and PassRole. - "policy affects public and cross-account access to your resource before deploying resource" (Access Analyzer's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html)) - "You can check whether the updated policy grants new access compared to the existing version." (Access Analyzer's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-custom-policy-checks.html)) ### One change, a class of risk: 1 of 10, weight 30 Access Analyzer only reads RCPs. No boundary generation is listed. - Limit: "Changes to a resource control policy (RCP) without a change to the bucket policy do not trigger a rescan of the bucket reported in the finding." (Access Analyzer's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-resources.html)) - Limit: "Identify external, internal, and unused access to your AWS resources." (Access Analyzer's own material, [read 10 Nov 2026](https://aws.amazon.com/iam/features/analyze-access/)) ### Uses the cloud's own controls: 4 of 10, weight 10 It generates native IAM policies from CloudTrail and is not in the request path. It works in one cloud only, and the user applies the result. - "IAM Access Analyzer generates IAM policies based on access activity in your AWS CloudTrail logs." (Access Analyzer's own material, [read 10 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html)) ### Stays fixed: 4 of 10, weight 10 Custom policy checks run before deployment in CI/CD and can trigger approval workflows. No drift or exception handling was found. - "Automate policy reviews before deployments by configuring custom policy checks in your development lifecycle." (Access Analyzer's own material, [read 10 Nov 2026](https://aws.amazon.com/iam/access-analyzer/)) - "The findings generated using IAM Access Analyzer's custom policy checks in our CI/CD pipeline triggers approval process workflows automatically." (Access Analyzer's own material, [read 10 Nov 2026](https://aws.amazon.com/iam/access-analyzer/)) ### Removes access, not only reports it: 3 of 10, weight 15 Refined policies and delete links are generated, and the user attaches or deletes them. - "For unused roles, access keys, or passwords, IAM Access Analyzer provides quick links in the console to help you delete them." (Access Analyzer's own material, [read 10 Nov 2026](https://aws.amazon.com/iam/access-analyzer/)) - Limit: "For unused permissions, IAM Access Analyzer reviews your existing policies and recommends a refined policy that is tailored to your access activity." (Access Analyzer's own material, [read 10 Nov 2026](https://aws.amazon.com/iam/access-analyzer/)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 4 | 1.20 | | One change, a class of risk | 30 | 1 | 0.30 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 2 | 0.04 | | Uses the cloud's own controls | 10 | 4 | 0.40 | | Evidence for auditors | 3 | 2 | 0.06 | Weighted total 2.9 (2.85 before rounding) on this site's weights. ## Access Analyzer in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 9th | 2.8 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 9th | 3.0 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 10th | 2.8 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 11th | 2.9 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 9th | 2.8 | Sonrai Security 5.5 | ## Strongest, in one sentence IAM Access Analyzer generates an IAM policy from up to 90 days of CloudTrail activity and recommends refined policies for unused permissions. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-policy-generation.html)) ## Related pages - [Wiz for guardrails: joint 8th of 16 at 2.9](/platforms/wiz/) - [P0 Security for guardrails: joint 10th of 16 at 2.3](/platforms/p0/) - [CrowdStrike for guardrails: joint 10th of 16 at 2.3](/platforms/crowdstrike/) ## Sources - aws.amazon.com/iam/access-analyzer: https://aws.amazon.com/iam/access-analyzer/ (read 10 Nov 2026) - docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 10 Nov 2026) - docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-custom-policy-checks.ht: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-custom-policy-checks.html (read 10 Nov 2026) - docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-policy-generation.html: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-policy-generation.html (read 10 Nov 2026) - docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-resources.html: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-resources.html (read 10 Nov 2026) - aws.amazon.com/iam/features/analyze-access: https://aws.amazon.com/iam/features/analyze-access/ (read 10 Nov 2026) - docs.aws.amazon.com/securityhub/latest/userguide/iam-controls.html: https://docs.aws.amazon.com/securityhub/latest/userguide/iam-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/aws-iam-access-analyzer/ --- # Wiz puts guardrails into production at 2.9, joint 8th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Wiz is joint 8th of 16 on how safely a platform puts a guardrail into production, at 2.9 of 10. It scores 4 on safe to enforce and 1 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Wiz sends a guardrail's fix to the person who owns the code. Wiz says it generates plans, assigns ownership and executes fixes, produces a pull request in the version control system, and applies security policies across IDEs, pull requests and CI/CD. SACR's report on Wiz asks whether Wiz's preventive model extends into cloud-native IAM, network, resource and organizational controls outside the pipeline, and says translating a requirement into durable controls across clouds is not yet validated. SACR adds that a generated fix is not an applied control until it reaches the customer's deployment process. Wiz's documented strength is one security graph that connects code, cloud, identities, network and runtime to prioritize exploitable risk and route fixes to code owners. For testing, Wiz uses Security Graph context and historical activity to judge whether a remediation could disrupt an application; SACR's example checked an access token's past use before deleting it. SACR adds that historical use cannot establish every infrequent or future dependency, and that attack-path validation is not a simulation of IAM, network or organizational policy changes. SACR lists exceptions, staged rollout, rollback, drift handling and cross-cloud consistency as needing further evidence for Wiz. ## How Wiz puts a guardrail into production ### Safe to enforce: 4 of 10, weight 30 Wiz evaluates historical activity to judge disruption before remediation. This is not a simulation of control changes, and infrequent dependencies are not covered. - "The Green Agent can use Security Graph context and historical activity to evaluate whether a proposed remediation could disrupt an underlying application." (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) - Limit: "although historical usage cannot establish every infrequent or future dependency." (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) ### One change, a class of risk: 1 of 10, weight 30 Only a limit is documented: translation into durable cross-cloud controls is unvalidated. - Limit: "SACR has not yet validated equivalent translation and enforcement depth across each provider, so this should be evaluated at the individual control and service level." (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) ### Uses the cloud's own controls: 2 of 10, weight 10 Wiz abstracts requirements from provider technology. Whether its output is a native preventive control is left open. - Limit: "whether the resulting control prevents the prohibited state or remediates it after creation." (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) - "Wiz described an approach centered on abstracting common security requirements from the underlying provider-specific technology." (softwareanalyst.io, counted as vendor material, [read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) ### Stays fixed: 4 of 10, weight 10 Policies across the IDE, pull requests and CI/CD can block unsafe changes, and Jira tickets are created. Exceptions, drift and paths outside the pipeline lack evidence. - Wiz generates a pull request with a fix in the version control system. (Wiz's own material, [read 10 Nov 2026](https://www.wiz.io/platform)) - "Apply consistent security policies across coding assistants, IDEs, version control, pull requests, and CI/CD pipelines, so developers catch issues early without changing how they work." (Wiz's own material, [read 10 Nov 2026](https://www.wiz.io/platform/wiz-code)) ### Removes access, not only reports it: 3 of 10, weight 15 Wiz generates plans, policies and pull requests sent to code owners. A generated fix is not an applied control. - "Automatically generate actionable plans, assign ownership, and execute fixes to drive your teams toward zero critical risk." (Wiz's own material, [read 10 Nov 2026](https://www.wiz.io)) - Limit: "Wiz analyzes cloud entitlements and auto-generates least privilege policies across your cloud, to help teams visualize, detect, prioritize, and remediate identity (IAM) risks." (Wiz's own material, [read 10 Nov 2026](https://www.wiz.io/solutions/ciem)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 4 | 1.20 | | One change, a class of risk | 30 | 1 | 0.30 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 6 | 0.12 | | Uses the cloud's own controls | 10 | 2 | 0.20 | | Evidence for auditors | 3 | 6 | 0.18 | Weighted total 2.9 (2.85 before rounding) on this site's weights. ## Wiz in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 8th | 3.0 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 8th | 3.1 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 7th | 3.5 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 7th | 3.4 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 8th | 3.0 | Sonrai Security 5.5 | ## Strongest, in one sentence Connects code, cloud, identities, network and runtime in one security graph to prioritize exploitable risk and route fixes to code owners. ([source](https://www.wiz.io)) ## Related pages - [P0 Security for guardrails: joint 10th of 16 at 2.3](/platforms/p0/) - [CrowdStrike for guardrails: joint 10th of 16 at 2.3](/platforms/crowdstrike/) - [Tenable Cloud Security for guardrails: joint 10th of 16 at 2.3](/platforms/tenable/) ## Sources - wiz.io: https://www.wiz.io (read 10 Nov 2026) - wiz.io/solutions/ciem: https://www.wiz.io/solutions/ciem (read 10 Nov 2026) - wiz.io/platform: https://www.wiz.io/platform (read 10 Nov 2026) - wiz.io/platform/wiz-code: https://www.wiz.io/platform/wiz-code (read 10 Nov 2026) - wiz.io/solutions/compliance: https://www.wiz.io/solutions/compliance (read 10 Nov 2026) - wiz.io/platform/wiz-cloud: https://www.wiz.io/platform/wiz-cloud (read 10 Nov 2026) - wiz.io/blog/introducing-azure-least-privilege-enforce-least-privilege-access-for: https://www.wiz.io/blog/introducing-azure-least-privilege-enforce-least-privilege-access-for-azure (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/wiz/ --- # P0 Security puts guardrails into production at 2.3, joint 10th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. P0 Security is joint 10th of 16 on how safely a platform puts a guardrail into production, at 2.3 of 10. It scores 0 on safe to enforce and 2 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. P0 has one deny that reads like a guardrail. In P0's Policy Studio documentation, a policy with a Deny all action for a production AWS account stops all access requests for that account through P0. It is a rule inside P0, not a cloud boundary. A P0 blog post says that in policy-based systems such as cloud providers, the constraints on P0's integration account can be implemented with conditional access management and permission boundaries. P0 provisions short-lived, scoped access and removes it when the work is done, with grant and revocation histories. For posture findings, P0 provides suggested shell commands that resolve a finding, and the page does not say P0 applies the fix. Nothing on simulating a change was found on 2026-11-10 after checking 20 pages. P0 policies can be managed as code with its Terraform provider, and findings can be filed as Jira issues. P0 connectors broker access from inside the customer's account. ## How P0 puts a guardrail into production ### Safe to enforce: 0 of 10, weight 30 None found on 2026-11-10 after checking 20 pages. None found on 10 Nov 2026 in the pages read for P0 Security. ### One change, a class of risk: 2 of 10, weight 30 A deny-all policy blocks access requests made through the product for an account. It is not a cloud boundary. - "Production (account `333333333333`): P0 denies all requests. No one can request access through P0." (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/policy-studio/policy-studio/configure-your-first-access-policy.md)) ### Uses the cloud's own controls: 5 of 10, weight 10 Default paths are agentless on native IAM, with session-bound roles in AWS and GCP. The AI Gateway sits in the agent data path, and connectors broker just-in-time access. - "P0's default access paths are agentless and driven entirely by your cloud provider's native IAM." (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/readme/connectors.md)) - "when P0 has an IAM integration with the upstream service (AWS and GCP today), enforcement happens inside that service via a session-bound temporary user, role, or policy" (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/readme/agentic-control-plane.md)) ### Stays fixed: 3 of 10, weight 10 Requests arrive through Slack, findings go to Jira, and a Terraform provider exists. No pre-production check or drift handling is described. - "any member of your organization can request access by typing `/p0 request` in Slack." (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/readme/just-in-time-access.md)) - "Install P0's Jira findings tracker for a Jira Cloud site, so P0 files findings as Jira issues as an Atlassian service account." (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/llms.txt)) ### Removes access, not only reports it: 5 of 10, weight 15 Just-in-time access is provisioned and revoked automatically, with history. Posture findings get suggested shell commands only. - "Provision only the privilege required for the task, then remove it when the work is done." (P0's own material, [read 10 Nov 2026](https://www.p0.dev/platform)) - "Access requests, together with their justifications, approval, grant, and revocation histories are recorded by P0, and can be exported to your SIEM." (P0's own material, [read 10 Nov 2026](https://docs.p0.dev/readme/just-in-time-access.md)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 0 | 0.00 | | One change, a class of risk | 30 | 2 | 0.60 | | Stays fixed | 10 | 3 | 0.30 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 5 | 0.50 | | Evidence for auditors | 3 | 3 | 0.09 | Weighted total 2.3 (2.32 before rounding) on this site's weights. ## P0 in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | joint 10th | 2.6 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 10th | 2.8 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 10th | 2.8 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 13th | 2.7 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 10th | 2.7 | Sonrai Security 5.5 | ## Strongest, in one sentence Grants short-lived, scoped access on request through Slack, CLI or web with approvals, and applies the same runtime authorization to AI agents. ([source](https://docs.p0.dev)) ## Related pages - [CrowdStrike for guardrails: joint 10th of 16 at 2.3](/platforms/crowdstrike/) - [Tenable Cloud Security for guardrails: joint 10th of 16 at 2.3](/platforms/tenable/) - [Entitle for guardrails: joint 13th of 16 at 2.0](/platforms/entitle/) ## Sources - p0.dev/platform: https://www.p0.dev/platform (read 10 Nov 2026) - docs.p0.dev/readme/just-in-time-access.md: https://docs.p0.dev/readme/just-in-time-access.md (read 10 Nov 2026) - docs.p0.dev/readme/posture.md: https://docs.p0.dev/readme/posture.md (read 10 Nov 2026) - docs.p0.dev/llms.txt: https://docs.p0.dev/llms.txt (read 10 Nov 2026) - docs.p0.dev/readme/access-inventory.md: https://docs.p0.dev/readme/access-inventory.md (read 10 Nov 2026) - docs.p0.dev: https://docs.p0.dev (read 10 Nov 2026) - docs.p0.dev/readme/connectors.md: https://docs.p0.dev/readme/connectors.md (read 10 Nov 2026) - docs.p0.dev/readme/agentic-control-plane.md: https://docs.p0.dev/readme/agentic-control-plane.md (read 10 Nov 2026) - p0.dev/customers: https://www.p0.dev/customers (read 10 Nov 2026) - docs.p0.dev/policy-studio/policy-studio/configure-your-first-access-policy.md: https://docs.p0.dev/policy-studio/policy-studio/configure-your-first-access-policy.md (read 10 Nov 2026) - p0.dev/blog/snowflake-least: https://p0.dev/blog/snowflake-least/ (read 10 Nov 2026) - docs.p0.dev/posture/finding-details.md: https://docs.p0.dev/posture/finding-details.md (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/p0/ --- # CrowdStrike puts guardrails into production at 2.3, joint 10th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. CrowdStrike is joint 10th of 16 on how safely a platform puts a guardrail into production, at 2.3 of 10. It scores 4 on safe to enforce and 0 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. CrowdStrike's Falcon Cloud Security generates policies and states a test for them. Its CIEM page says it generates least-privilege access policies, and its product page says risky user access is shut down with resource-based policies. CrowdStrike's announcement lists remediation testing that simulates a remediation as a dry run before deployment, without saying what the simulation runs against. A policy covering a class of risk was not found: none found on 2026-11-10 after checking 16 pages. CrowdStrike's Falcon scans infrastructure as code in CI/CD, IDEs and MCP workflows before deployment, and Kubernetes admission controls block flagged container images before they reach production. CrowdStrike's explainer says CIEM tools generate remediation recommendations, and it does not state that Falcon applies the change. No native enforcement mechanism was found for Falcon Cloud Security on 2026-11-10 after checking 17 pages. CrowdStrike's documented strength is that it ties cloud findings to adversary tradecraft it has observed to rank which exposures to fix first. ## How CrowdStrike puts a guardrail into production ### Safe to enforce: 4 of 10, weight 30 CrowdStrike states a dry-run simulation of remediation before deployment, without saying what it runs against. - "One-click remediation testing: Simulate remediation tactics to understand outcomes and ensure confidence by performing a dry run prior to deployment." (CrowdStrike's own material, [read 10 Nov 2026](https://www.crowdstrike.com/blog/crowdstrike-extends-ciem-capabilities/)) ### One change, a class of risk: 0 of 10, weight 30 None found on 2026-11-10 after checking 16 pages. None found on 10 Nov 2026 in the pages read for CrowdStrike. ### Uses the cloud's own controls: 0 of 10, weight 10 None found on 2026-11-10 after checking 17 pages. None found on 10 Nov 2026 in the pages read for CrowdStrike. ### Stays fixed: 4 of 10, weight 10 Infrastructure-as-code scanning runs in CI/CD, the IDE and MCP, and Kubernetes admission can block. No exception or drift handling was found. - "Scan infrastructure as code in CI/CD, IDEs, and MCP workflows so developers stop misconfigurations before deployment." (CrowdStrike's own material, [read 10 Nov 2026](https://www.crowdstrike.com/en-us/platform/cloud-security/cspm/)) - "Kubernetes admission controls block affected images before they reach production." (CrowdStrike's own material, [read 10 Nov 2026](https://www.crowdstrike.com/en-us/platform/cloud-security/)) ### Removes access, not only reports it: 3 of 10, weight 15 CrowdStrike generates least-privilege and resource-based policies and says it removes unwanted access, but recommendations are what is described as output. - "Analyze cloud permissions and generate least-privilege access policies to prevent lateral movement." (CrowdStrike's own material, [read 10 Nov 2026](https://www.crowdstrike.com/en-us/platform/cloud-security/cnapp/)) - Limit: "generate remediation recommendations" (CrowdStrike's own material, [read 10 Nov 2026](https://www.crowdstrike.com/en-us/cybersecurity-101/cloud-security/cloud-infrastructure-entitlement-management-ciem/)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 4 | 1.20 | | One change, a class of risk | 30 | 0 | 0.00 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 0 | 0.00 | | Evidence for auditors | 3 | 6 | 0.18 | Weighted total 2.3 (2.31 before rounding) on this site's weights. ## CrowdStrike in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 12th | 2.5 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 12th | 2.5 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 10th | 2.8 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 9th | 3.0 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 12th | 2.5 | Sonrai Security 5.5 | ## Strongest, in one sentence CrowdStrike ties cloud findings to adversary tradecraft it has observed to rank which exposures to fix first. ([source](https://www.crowdstrike.com/en-us/platform/cloud-security/cspm/)) ## Related pages - [Tenable Cloud Security for guardrails: joint 10th of 16 at 2.3](/platforms/tenable/) - [Entitle for guardrails: joint 13th of 16 at 2.0](/platforms/entitle/) - [Sysdig for guardrails: joint 13th of 16 at 2.0](/platforms/sysdig/) ## Sources - crowdstrike.com/en-us/platform/cloud-security/cnapp: https://www.crowdstrike.com/en-us/platform/cloud-security/cnapp/ (read 10 Nov 2026) - crowdstrike.com/en-us/cybersecurity-101/cloud-security/cloud-infrastructure-enti: https://www.crowdstrike.com/en-us/cybersecurity-101/cloud-security/cloud-infrastructure-entitlement-management-ciem/ (read 10 Nov 2026) - crowdstrike.com/en-us/platform/cloud-security/ciem: https://www.crowdstrike.com/en-us/platform/cloud-security/ciem/ (read 10 Nov 2026) - crowdstrike.com/en-us/platform/cloud-security/cspm: https://www.crowdstrike.com/en-us/platform/cloud-security/cspm/ (read 10 Nov 2026) - crowdstrike.com/en-us/platform/cloud-security: https://www.crowdstrike.com/en-us/platform/cloud-security/ (read 10 Nov 2026) - crowdstrike.com/en-us/platform/cloud-security/ai-spm: https://www.crowdstrike.com/en-us/platform/cloud-security/ai-spm/ (read 10 Nov 2026) - crowdstrike.com/en-us/platform/cloud-security/cloud-compliance: https://www.crowdstrike.com/en-us/platform/cloud-security/cloud-compliance/ (read 10 Nov 2026) - crowdstrike.com/blog/crowdstrike-extends-ciem-capabilities: https://www.crowdstrike.com/blog/crowdstrike-extends-ciem-capabilities/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/crowdstrike/ --- # Tenable Cloud Security puts guardrails into production at 2.3, joint 10th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Tenable Cloud Security is joint 10th of 16 on how safely a platform puts a guardrail into production, at 2.3 of 10. It scores 3 on safe to enforce and 0 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Tenable Cloud Security writes policies per identity. Tenable says its CIEM engine identifies excessive entitlements and enforces least-privilege policies, and its blog describes resource-level policies that the user can apply. A guardrail that covers a class of risk, such as an organization-level deny, was not found in Tenable's pages: none found on 2026-11-10 after checking 19 pages. Tenable's JIT access grants time-limited permissions in place of standing privilege. The only preview in Tenable's evidence is the diff the product shows between a generated policy and the current policy; no simulation against the environment is described. Tenable scans infrastructure-as-code templates (Terraform, CloudFormation, Kubernetes manifests) for violations, and sends tickets to Jira, Slack and Microsoft Teams. Whether the product itself or the user applies a remediation is not stated on the pages read, and no native enforcement mechanism was found for Tenable on 2026-11-10. ## How Tenable puts a guardrail into production ### Safe to enforce: 3 of 10, weight 30 The product shows the diff between a generated policy and the current one. No simulation against the environment is described. - Tenable's blog says the product generates right-sized policies the user can apply and shows the exact difference between the generated policy and the current policy. The page does not describe a simulation against the environment. (Tenable's own material, [read 10 Nov 2026](https://www.tenable.com/blog/least-privilege-policy-automated-analysis-trumps-native-aws-tools)) ### One change, a class of risk: 0 of 10, weight 30 None found on 2026-11-10 after checking 19 pages. None found on 10 Nov 2026 in the pages read for Tenable Cloud Security. ### Uses the cloud's own controls: 0 of 10, weight 10 None found on 2026-11-10 after checking 19 pages. None found on 10 Nov 2026 in the pages read for Tenable Cloud Security. ### Stays fixed: 4 of 10, weight 10 Infrastructure-as-code template scanning is offered, with tickets and approvals in Jira, Slack and Teams. No exception or drift handling is described. - IaC templates (Terraform, CloudFormation, Kubernetes manifests) are scanned for misconfigurations and policy violations. (Tenable's own material, [read 10 Nov 2026](https://www.tenable.com/products/tenable-cloud-security)) - "utilize standard communication tools such as Jira, Slack, Microsoft Teams, and email integration tools to scale the tool within your organization." (Tenable's own material, [read 10 Nov 2026](https://www.tenable.com/products/tenable-cloud-security)) ### Removes access, not only reports it: 5 of 10, weight 15 Time-limited just-in-time permissions replace standing privilege. Right-sizing policies are applied by the user, and remediation workflows do not say who applies them. - "Use the Tenable cloud infrastructure and entitlements management (CIEM) engine to identify excessive or toxic entitlements and enforce least-privilege policies." (Tenable's own material, [read 10 Nov 2026](https://www.tenable.com/products/tenable-cloud-security)) - "Automated remediation workflows help security teams right-size permissions at scale without disrupting development velocity." (Tenable's own material, [read 10 Nov 2026](https://www.tenable.com/products/tenable-cloud-security)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 3 | 0.90 | | One change, a class of risk | 30 | 0 | 0.00 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 5 | 0.10 | | Uses the cloud's own controls | 10 | 0 | 0.00 | | Evidence for auditors | 3 | 5 | 0.15 | Weighted total 2.3 (2.30 before rounding) on this site's weights. ## Tenable in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | joint 10th | 2.6 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 11th | 2.7 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 9th | 3.1 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 9th | 3.0 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 10th | 2.7 | Sonrai Security 5.5 | ## Strongest, in one sentence Combines CIEM entitlement analysis and JIT access with posture management, IaC scanning and attack paths in one cloud platform. ([source](https://www.tenable.com/cloud-security)) ## Related pages - [Entitle for guardrails: joint 13th of 16 at 2.0](/platforms/entitle/) - [Sysdig for guardrails: joint 13th of 16 at 2.0](/platforms/sysdig/) - [Prisma Cloud for guardrails: 15th of 16 at 1.7](/platforms/prisma/) ## Sources - tenable.com/products/tenable-cloud-security: https://www.tenable.com/products/tenable-cloud-security (read 10 Nov 2026) - tenable.com/cloud-security: https://www.tenable.com/cloud-security (read 10 Nov 2026) - tenable.com/blog/least-privilege-policy-automated-analysis-trumps-native-aws-too: https://www.tenable.com/blog/least-privilege-policy-automated-analysis-trumps-native-aws-tools (read 10 Nov 2026) - tenable.com/blog/the-future-of-cloud-access-management-how-tenable-cloud-securit: https://www.tenable.com/blog/the-future-of-cloud-access-management-how-tenable-cloud-security-redefines-just-in-time-access (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/tenable/ --- # Entitle puts guardrails into production at 2.0, joint 13th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Entitle is joint 13th of 16 on how safely a platform puts a guardrail into production, at 2.0 of 10. It scores 0 on safe to enforce and 2 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. A cloud guardrail in Entitle's pages: none found on 2026-11-10. Entitle's integration rules apply a setting across many resources or roles by condition, including ones created later, but the settings listed are approval workflow, owner and requestable status. The documentation lists no deny or boundary setting, so Entitle governs how access is requested rather than what a policy blocks. Entitle's birthright policies grant a defined set of permissions to a group and remove them when the person leaves it. Entitle provisions access through its own agent, installed in the customer VPC with Terraform and Helm charts, rather than through a cloud policy object. Entitle auto-revokes elevated access after a set duration, ticket resolution or on-call rotation. On safe rollout, nothing on simulating or previewing the effect of a change was found on 2026-11-10 after checking 15 pages. Entitle offers a Terraform provider and REST APIs, and approved requests can be processed through a ticketing system. ## How Entitle puts a guardrail into production ### Safe to enforce: 0 of 10, weight 30 None found on 2026-11-10 after checking 15 pages. None found on 10 Nov 2026 in the pages read for Entitle. ### One change, a class of risk: 2 of 10, weight 30 Birthright policies and access policies apply request settings across many resources. There is no deny or boundary. - "Add rules to apply a policy across many resources or roles at once, and keep new resources and roles governed automatically as they appear" (Entitle's own material, [read 10 Nov 2026](https://docs.beyondtrust.com/entitle/docs/rules)) - Limit: "Apply an approval workflow to resources or roles automatically Set an owner or maintainer for resources Make resources or roles requestable" (Entitle's own material, [read 10 Nov 2026](https://docs.beyondtrust.com/entitle/docs/rules)) ### Uses the cloud's own controls: 2 of 10, weight 10 Access is provisioned through the vendor's own VPC agent, which has outbound access, and an API sidecar. - "First, you install an agent on your virtual private cloud (VPC) using Terraform and Helms Charts." (Entitle's own material, [read 10 Nov 2026](https://www.beyondtrust.com/products/entitle)) - Limit: "The Entitle agent requires outbound network access to function." (Entitle's own material, [read 10 Nov 2026](https://docs.beyondtrust.com/entitle/docs/entitle-agent-network-access-requirements)) ### Stays fixed: 3 of 10, weight 10 Ticketing fulfilment, a Terraform provider and REST APIs are offered. No pre-production check or drift handling is described. - "Third, we offer manual fulfillment through a ticketing system where approved access requests are processed by your IT service team." (Entitle's own material, [read 10 Nov 2026](https://www.beyondtrust.com/products/entitle)) - "Customize to fit your exact needs using REST APIs and a Terraform Provider." (Entitle's own material, [read 10 Nov 2026](https://www.beyondtrust.com/products/entitle)) ### Removes access, not only reports it: 5 of 10, weight 15 Elevated access is revoked automatically after a set duration, a ticket or a rotation, and approvers can take bulk actions on identities. - "Enable controlled, self-serving privileged access that is auto revoked after a set duration, ticket resolution, or on-call rotation." (entitle.io, counted as vendor material, [read 10 Nov 2026](https://www.entitle.io)) - "Reports are automatically generated so you can review entitlements and take bulk actions on identities across your IaaS and SaaS environments." (Entitle's own material, [read 10 Nov 2026](https://www.beyondtrust.com/products/entitle)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 5 | 0.75 | | Safe to enforce | 30 | 0 | 0.00 | | One change, a class of risk | 30 | 2 | 0.60 | | Stays fixed | 10 | 3 | 0.30 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 2 | 0.20 | | Evidence for auditors | 3 | 3 | 0.09 | Weighted total 2.0 (2.02 before rounding) on this site's weights. ## Entitle in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 13th | 2.3 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | joint 13th | 2.4 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 13th | 2.6 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 14th | 2.4 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | joint 12th | 2.5 | Sonrai Security 5.5 | ## Strongest, in one sentence Entitle provides just-in-time cloud and SaaS access that auto-revokes, with approval workflows in Slack and Teams, access reviews and audit logs. ([source](https://www.entitle.io)) ## Related pages - [Sysdig for guardrails: joint 13th of 16 at 2.0](/platforms/sysdig/) - [Prisma Cloud for guardrails: 15th of 16 at 1.7](/platforms/prisma/) - [Microsoft for guardrails: 16th of 16 at 1.1](/platforms/microsoft/) ## Sources - entitle.io: https://www.entitle.io (read 10 Nov 2026) - beyondtrust.com/products/entitle: https://www.beyondtrust.com/products/entitle (read 10 Nov 2026) - docs.beyondtrust.com/entitle/docs/entitle-agent-network-access-requirements: https://docs.beyondtrust.com/entitle/docs/entitle-agent-network-access-requirements (read 10 Nov 2026) - entitle.io/security: https://www.entitle.io/security (read 10 Nov 2026) - docs.beyondtrust.com/entitle/docs/rules: https://docs.beyondtrust.com/entitle/docs/rules (read 10 Nov 2026) - docs.beyondtrust.com/entitle/docs/birthright-policies: https://docs.beyondtrust.com/entitle/docs/birthright-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/entitle/ --- # Sysdig puts guardrails into production at 2.0, joint 13th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Sysdig is joint 13th of 16 on how safely a platform puts a guardrail into production, at 2.0 of 10. It scores 2 on safe to enforce and 0 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Sysdig turns runtime observation into a policy for the user to apply. Sysdig's CIEM page says it analyzes access patterns and generates access policies that restrict privileges to what is used, and its blog says to validate the suggested policy before applying it. Sysdig does not say it applies the policy to the account. A class-level guardrail, such as an organization deny, was not found: none found on 2026-11-10 after checking 17 pages. Sysdig says its recommendations are meant to tighten IAM policies without disrupting legitimate activity, and no simulation step is described. For AI agents, Sysdig AI Defense starts policies in alert mode before enforcement, and blocks agent actions before they run using kernel-level instrumentation, so that enforcement is Sysdig software on the host. Sysdig's infrastructure-as-code scanner checks files at commit, pull request and CI/CD stages and can fail builds. ## How Sysdig puts a guardrail into production ### Safe to enforce: 2 of 10, weight 30 Recommendations are claimed to be non-disruptive, and AI agent policies start in alert mode. No simulation step is described. - "With Sysdig’s careful recommendations, teams can confidently tighten IAM policies without disrupting legitimate activity." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) - "Policies begin in alert mode and move to enforcement when a team is ready." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/press-releases/sysdig-launches-sysdig-ai-defense-bringing-runtime-security-to-every-action-ai-agents-take)) ### One change, a class of risk: 0 of 10, weight 30 None found on 2026-11-10 after checking 17 pages. None found on 10 Nov 2026 in the pages read for Sysdig. ### Uses the cloud's own controls: 3 of 10, weight 10 For cloud, the output is a generated IAM policy that the user applies. Agent enforcement uses the product's own kernel-level host software. - "It checks every action against the organization’s policy, blocks the ones that break it before they run, and kills the process if one gets past." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/press-releases/sysdig-launches-sysdig-ai-defense-bringing-runtime-security-to-every-action-ai-agents-take)) - "Optimize permissions with automatically generated policies and total visibility of in-use permissions" (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) ### Stays fixed: 4 of 10, weight 10 Infrastructure-as-code scanning at commit, pull request and CI/CD can fail builds. No exception handling in the team's tools and no drift handling were found. - "These checks run automatically during commits, pull requests, and CI/CD pipelines, and then the system provides actionable feedback with remediation guidance." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/infrastructure-as-code-security)) - "It will fail builds when critical issues are detected and provide detailed remediation guidance to developers." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/infrastructure-as-code-security)) ### Removes access, not only reports it: 3 of 10, weight 15 Sysdig generates least-privilege policies and infrastructure-as-code fix pull requests. The user validates and applies them. - Limit: "receive suggested least-privilege versions of existing policies, making cleanup fast and low risk." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) - "Sysdig analyzes access patterns to identify unused permissions and automatically generate access policies that restrict privileges to precisely what users need to perform their jobs effectively." (Sysdig's own material, [read 10 Nov 2026](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 2 | 0.60 | | One change, a class of risk | 30 | 0 | 0.00 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 4 | 0.08 | | Uses the cloud's own controls | 10 | 3 | 0.30 | | Evidence for auditors | 3 | 6 | 0.18 | Weighted total 2.0 (2.01 before rounding) on this site's weights. ## Sysdig in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 14th | 2.2 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | joint 13th | 2.4 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | joint 13th | 2.6 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 12th | 2.8 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 14th | 2.2 | Sonrai Security 5.5 | ## Strongest, in one sentence Sysdig uses runtime data to show which permissions are in use and generates access policies limited to that use. ([source](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) ## Related pages - [Prisma Cloud for guardrails: 15th of 16 at 1.7](/platforms/prisma/) - [Microsoft for guardrails: 16th of 16 at 1.1](/platforms/microsoft/) - [Act Security for guardrails: 1st of 16 at 5.9](/platforms/act/) ## Sources - sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem: https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem (read 10 Nov 2026) - sysdig.com/about: https://www.sysdig.com/about (read 10 Nov 2026) - sysdig.com/press-releases/sysdig-launches-sysdig-ai-defense-bringing-runtime-sec: https://www.sysdig.com/press-releases/sysdig-launches-sysdig-ai-defense-bringing-runtime-security-to-every-action-ai-agents-take (read 10 Nov 2026) - sysdig.com/solutions/infrastructure-as-code-security: https://www.sysdig.com/solutions/infrastructure-as-code-security (read 10 Nov 2026) - sysdig.com/products/secure: https://www.sysdig.com/products/secure (read 10 Nov 2026) - sysdig.com/solutions/cloud-security-posture-management-cspm: https://www.sysdig.com/solutions/cloud-security-posture-management-cspm (read 10 Nov 2026) - sysdig.com/blog/ciem-security-sysdig-secure: https://www.sysdig.com/blog/ciem-security-sysdig-secure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/sysdig/ --- # Prisma Cloud puts guardrails into production at 1.7, 15th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Prisma Cloud is 15th of 16 on how safely a platform puts a guardrail into production, at 1.7 of 10. It scores 0 on safe to enforce and 1 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Prisma Cloud reads guardrails as input: its effective-permission calculation takes SCPs into account, and a feature that creates organization-level guardrails: none found on 2026-11-10. IAM remediation is a CLI command the user runs, or a custom script the user configures, and Prisma's documentation says automatic remediation of GCP IAM alerts is not supported. Prisma's next version, marketed as Cortex Cloud, says it uses AI-driven analysis to remove unused entitlements. Prisma Cloud's pre-production check is in the pipeline: pull request scans fail the PR at a severity threshold and add PR comments. Alerts go to ServiceNow and Slack, among 14 tools. Nothing on simulating a change's effect was found on 2026-11-10 after checking 19 pages. Prisma's remediation runs native CLI commands in the customer's own environment, with no proxy in the request path, and the user executes them. ## How Prisma Cloud puts a guardrail into production ### Safe to enforce: 0 of 10, weight 30 None found on 2026-11-10 after checking 19 pages. None found on 10 Nov 2026 in the pages read for Prisma Cloud. ### One change, a class of risk: 1 of 10, weight 30 Prisma Cloud only reads SCPs as input. It does not create organization-level guardrails. - Limit: "the algorithm combines various cloud sources such as AWS IAM roles, AWS IAM policies, AWS IAM groups, AWS resource based policies, and AWS service control policies (SCPs) to compute the net effective permissions" (docs.prismacloud.io, counted as vendor material, [read 10 Nov 2026](https://docs.prismacloud.io/content-collections/administration/configure-iam-security/what-is-prisma-cloud-iam-security.md)) ### Uses the cloud's own controls: 3 of 10, weight 10 Remediation runs as native CLI commands that the user executes, with no proxy. It is manual. - Limit: IAM remediation runs CLI commands (or a script) in the customer's own cloud environment; no proxy in the request path was described on pages read. (docs.prismacloud.io, counted as vendor material, [read 10 Nov 2026](https://docs.prismacloud.io/content-collections/administration/configure-iam-security/remediate-alerts-for-iam-security.md)) ### Stays fixed: 4 of 10, weight 10 Pull request scans fail at a threshold, alerts go to ServiceNow and Slack, and requests run through ServiceNow. No drift or exception handling was found. - "the PR scan will fail, and PR comments will be added." (docs.prismacloud.io, counted as vendor material, [read 10 Nov 2026](https://docs.prismacloud.io/content-collections/application-security/risk-management/monitor-and-manage-code-build/pull-request-scan.md)) - "Send alert notifications to 14 third-party tools, including email, AWS Lambda and Security Hub, PagerDuty®, ServiceNow® and Slack®." (Prisma Cloud's own material, [read 10 Nov 2026](https://www.paloaltonetworks.com/prisma/cloud/cloud-infrastructure-entitlement-management)) ### Removes access, not only reports it: 3 of 10, weight 15 IAM fixes are CLI commands the user runs, or a script the user configures. There is no GCP auto-remediation. - Limit: "You can manually remediate the alerts by copying the AWS, Azure, or GCP CLI commands and then run them in your cloud environment or you can configure a custom python script to automate the remediation steps." (docs.prismacloud.io, counted as vendor material, [read 10 Nov 2026](https://docs.prismacloud.io/content-collections/administration/configure-iam-security/remediate-alerts-for-iam-security.md)) - Limit: "Automatic remediation for GCP IAM alerts is not supported." (docs.prismacloud.io, counted as vendor material, [read 10 Nov 2026](https://docs.prismacloud.io/content-collections/administration/configure-iam-security/remediate-alerts-for-iam-security.md)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 3 | 0.45 | | Safe to enforce | 30 | 0 | 0.00 | | One change, a class of risk | 30 | 1 | 0.30 | | Stays fixed | 10 | 4 | 0.40 | | Identity, network and AI access together | 2 | 5 | 0.10 | | Uses the cloud's own controls | 10 | 3 | 0.30 | | Evidence for auditors | 3 | 5 | 0.15 | Weighted total 1.7 (1.70 before rounding) on this site's weights. ## Prisma Cloud in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 15th | 2.0 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 15th | 2.2 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 15th | 2.5 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | joint 14th | 2.4 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 15th | 2.1 | Sonrai Security 5.5 | ## Strongest, in one sentence Calculates net effective permissions across clouds, including SCPs as input, and ties them to a broad platform with code scanning, network analysis and a large library of compliance standards. ([source](https://docs.prismacloud.io/content-collections/administration/configure-iam-security/what-is-prisma-cloud-iam-security.md)) ## Related pages - [Microsoft for guardrails: 16th of 16 at 1.1](/platforms/microsoft/) - [Act Security for guardrails: 1st of 16 at 5.9](/platforms/act/) - [Sonrai Security for guardrails: 2nd of 16 at 5.7](/platforms/sonrai/) ## Sources - docs.prismacloud.io/content-collections/administration/configure-iam-security/re: https://docs.prismacloud.io/content-collections/administration/configure-iam-security/remediate-alerts-for-iam-security.md (read 10 Nov 2026) - paloaltonetworks.com/cortex/cloud/cloud-infrastructure-entitlement-management: https://www.paloaltonetworks.com/cortex/cloud/cloud-infrastructure-entitlement-management (read 10 Nov 2026) - docs.prismacloud.io/content-collections/administration/configure-iam-security/wh: https://docs.prismacloud.io/content-collections/administration/configure-iam-security/what-is-prisma-cloud-iam-security.md (read 10 Nov 2026) - docs.prismacloud.io/content-collections/application-security/risk-management/mon: https://docs.prismacloud.io/content-collections/application-security/risk-management/monitor-and-manage-code-build/pull-request-scan.md (read 10 Nov 2026) - paloaltonetworks.com/prisma/cloud/cloud-infrastructure-entitlement-management: https://www.paloaltonetworks.com/prisma/cloud/cloud-infrastructure-entitlement-management (read 10 Nov 2026) - docs.prismacloud.io/content-collections/administration/network-security/cloud-ne: https://docs.prismacloud.io/content-collections/administration/network-security/cloud-network-analyzer.md (read 10 Nov 2026) - docs.prismacloud.io/content-collections/compliance/compliance-standards.md: https://docs.prismacloud.io/content-collections/compliance/compliance-standards.md (read 10 Nov 2026) - paloaltonetworks.com/customers/michaels-adopts-least-privilege-access-strategy-w: https://www.paloaltonetworks.com/customers/michaels-adopts-least-privilege-access-strategy-with-prisma-cloud-ciem (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/prisma/ --- # Microsoft puts guardrails into production at 1.1, 16th of 16 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Microsoft is 16th of 16 on how safely a platform puts a guardrail into production, at 1.1 of 10. It scores 0 on safe to enforce and 1 on one change for a class of risk, the two criteria weighted 30 each here. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. Microsoft is scored here on cloud infrastructure entitlement management in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. Defender for Cloud gives recommendations such as right-sizing permissions, described by Microsoft as guidance for remediation, and from 2026-08-06 it no longer publishes unused-action details for AWS. The retired Entra Permissions Management documentation describes a Remediation dashboard where an administrator authors a role or policy with an Allow or Deny effect in JSON, and a Preview page that shows the generated script for review. That preview is a script review, not a simulation of impact. Azure Policy is a separate Azure service that assigns definitions to management groups, subscriptions or resource groups, and Microsoft's Defender for Cloud pages do not describe it acting on CIEM findings; the Azure Policy mechanisms are covered in this library's Azure section. Defender for Cloud correlates infrastructure-as-code misconfigurations with cloud context, and the Entra Permissions Management FAQ listed ITSM integration as roadmap. Microsoft's documented strength is that Defender for Cloud evaluates effective permissions to identify which identities can reach sensitive or business-critical resources. ## How Microsoft puts a guardrail into production ### Safe to enforce: 0 of 10, weight 30 Microsoft is scored on cloud infrastructure entitlement management (CIEM) in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. None found on 2026-11-10 for Defender for Cloud. - Limit: "Select Mode , and then, if you want recommendations to be generated and applied manually, select On-Demand ." (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/entra/permissions-management/how-to-create-rule)) ### One change, a class of risk: 1 of 10, weight 30 Microsoft is scored on cloud infrastructure entitlement management (CIEM) in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. What remains is Azure Policy, a separate Azure service assigned to management groups and not described as acting on CIEM findings. - Limit: "Role Last Used On Is : Select the duration in days when the role was last used." (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/entra/permissions-management/how-to-create-rule)) - "the policy definition or initiative is assigned to any scope of resources that Azure supports. For example, management groups" (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/azure/governance/policy/overview)) ### Uses the cloud's own controls: 0 of 10, weight 10 Microsoft is scored on cloud infrastructure entitlement management (CIEM) in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. None found on 2026-11-10 for Defender for Cloud. None found on 10 Nov 2026 in the pages read for Microsoft. ### Stays fixed: 2 of 10, weight 10 Only IaC misconfiguration correlation is documented, and ITSM integration was presented as roadmap. - Limit: "Integration with ITSM tools, such as ServiceNow, is in the future roadmap." (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/entra/permissions-management/faqs)) - "DevOps security findings, such as Infrastructure as Code (IaC) misconfigurations and exposed secrets, can then be correlated with other contextual cloud security insights to prioritize remediation" (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/azure/defender-for-cloud/defender-for-cloud-introduction)) ### Removes access, not only reports it: 2 of 10, weight 15 Microsoft is scored on cloud infrastructure entitlement management (CIEM) in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. Defender for Cloud CIEM gives recommendations described as guidance for remediating, and from 2026-08-06 no longer publishes unused-action details for AWS. - Limit: "Right-sizing permissions for overprovisioned identities based on actual usage" (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/azure/defender-for-cloud/permissions-management)) - Limit: "The built-in recommendations provide guidance for remediating inactive identities, overprovisioned permissions, and insecure identity settings." (Microsoft's own material, [read 10 Nov 2026](https://learn.microsoft.com/en-us/azure/defender-for-cloud/permissions-management)) ## All seven scores | Criterion | Weight here | Score | Weighted points | | --- | --- | --- | --- | | Removes access, not only reports it | 15 | 2 | 0.30 | | Safe to enforce | 30 | 0 | 0.00 | | One change, a class of risk | 30 | 1 | 0.30 | | Stays fixed | 10 | 2 | 0.20 | | Identity, network and AI access together | 2 | 6 | 0.12 | | Uses the cloud's own controls | 10 | 0 | 0.00 | | Evidence for auditors | 3 | 5 | 0.15 | Weighted total 1.1 (1.07 before rounding) on this site's weights. ## Microsoft in the five situations | Situation | Position | Score | Leader | | --- | --- | --- | --- | | [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/) | 16th | 1.3 | Act Security 5.6 | | [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/) | 16th | 1.5 | Act Security 5.6 | | [A company putting AI agents into production this year](/situations/ai-agents/) | 16th | 2.1 | Act Security 5.6 | | [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/) | 16th | 1.8 | Sonrai Security 5.2 | | [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/) | 16th | 1.4 | Sonrai Security 5.5 | ## Strongest, in one sentence Defender for Cloud evaluates effective permissions to identify which identities can reach sensitive or business-critical resources. ([source](https://learn.microsoft.com/en-us/azure/defender-for-cloud/permissions-management)) ## Related pages - [Act Security for guardrails: 1st of 16 at 5.9](/platforms/act/) - [Sonrai Security for guardrails: 2nd of 16 at 5.7](/platforms/sonrai/) - [AWS Control Tower for guardrails: 3rd of 16 at 4.2](/platforms/aws-control-tower/) ## Sources - learn.microsoft.com/en-us/azure/defender-for-cloud/permissions-management: https://learn.microsoft.com/en-us/azure/defender-for-cloud/permissions-management (read 10 Nov 2026) - learn.microsoft.com/en-us/entra/permissions-management/faqs: https://learn.microsoft.com/en-us/entra/permissions-management/faqs (read 10 Nov 2026) - learn.microsoft.com/en-us/azure/defender-for-cloud/defender-for-cloud-introducti: https://learn.microsoft.com/en-us/azure/defender-for-cloud/defender-for-cloud-introduction (read 10 Nov 2026) - learn.microsoft.com/en-us/azure/defender-for-cloud/concept-attack-path: https://learn.microsoft.com/en-us/azure/defender-for-cloud/concept-attack-path (read 10 Nov 2026) - learn.microsoft.com/en-us/azure/defender-for-cloud/regulatory-compliance-dashboa: https://learn.microsoft.com/en-us/azure/defender-for-cloud/regulatory-compliance-dashboard (read 10 Nov 2026) - learn.microsoft.com/en-us/azure/defender-for-cloud/enable-permissions-management: https://learn.microsoft.com/en-us/azure/defender-for-cloud/enable-permissions-management (read 10 Nov 2026) - learn.microsoft.com/en-us/azure/governance/policy/overview: https://learn.microsoft.com/en-us/azure/governance/policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/platforms/microsoft/ --- # Five situations, five re-weighted rankings Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Five buyer situations re-weight the same sixteen platforms: hundreds of AWS accounts, several clouds, a regulated company, AI agents in production and a small team. Act is first in three and shares first with Sonrai Security in two, where Sonrai is listed first. - [An AWS-heavy organization with hundreds of accounts and years of accumulated access](/situations/aws-scale/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A multi-cloud enterprise with separate teams per cloud](/situations/multi-cloud/): First: Act Security 5.6; then Sonrai Security 5.5; Act 1st - [A company putting AI agents into production this year](/situations/ai-agents/): First: Act Security 5.6; then Sonrai Security 5.2; Act 1st - [A regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53](/situations/regulated/): First: Sonrai Security 5.2 (joint with Act Security); Act joint 1st - [A small cloud security team buried under posture findings it cannot clear](/situations/small-team/): First: Sonrai Security 5.5 (joint with Act Security); Act joint 1st **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Guides](/guides/) - [Glossary](/glossary/) - [News](/news/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/ --- # For an AWS estate of hundreds of accounts, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.3 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. For an AWS-heavy organization with hundreds of accounts and years of accumulated access, the top three on this situation's weights are Act Security (5.6), Sonrai Security (5.5) and AWS Control Tower (4.3). Hundreds of accounts make permission-by-permission clean-up impossible; one change must remove a class of paths, safely. Act is first at 5.6, ahead of Sonrai Security at 5.5. At hundreds of accounts a guardrail has to be one policy attached high in the organization, because editing permissions account by account does not finish. This situation weights one change for a class of risk at 27.5, safe to enforce at 25 and removes at 20, with stays fixed at 12.5. The weights say what an estate of this size needs: a service control policy or resource control policy that covers an OU, shown safe before it lands on the root. The ranking rewards products that write that policy in AWS's own mechanisms and show its effect first. Two limits of the mechanisms matter at scale. AWS documents a 10,240 character ceiling on an SCP and at most 10 SCPs attached directly to a root, OU or account, and says SCPs do not affect users or roles in the management account. A platform that plans around those limits earns its place in this view. ## The top three for this situation ### Rank 1: Act Security (5.6) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 2: Sonrai Security (5.5) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 3: AWS Control Tower (4.3) Provider guardrail service, One change, a class of risk 6 Highest score: one change, a class of risk, 6 of 10. Lowest: safe to enforce, 1. [Read the profile](/platforms/aws-control-tower/) ## The full order under these weights | Rank | Platform | Total | Removes (20) | Safe (25) | Class (27.5) | Fixed (12.5) | Id+Net+AI (3.5) | Native (7.5) | Audit (4) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.6 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.5 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.3 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Turbot](/platforms/turbot/) | 4.1 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 5 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.0 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 6 | [Apono](/platforms/apono/) | 3.6 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.1 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8 | [Wiz](/platforms/wiz/) | 3.0 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 9 | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.8 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 10= | [Tenable Cloud Security](/platforms/tenable/) | 2.6 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 10= | [P0 Security](/platforms/p0/) | 2.6 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 12 | [CrowdStrike](/platforms/crowdstrike/) | 2.5 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 13 | [Entitle](/platforms/entitle/) | 2.3 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 14 | [Sysdig](/platforms/sysdig/) | 2.2 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 2.0 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.3 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Published weights for this situation, the mean of this site's weights and the situation's own (removes 25, safe 20, class 25, fixed 15, id+net+ai 5, native 5, audit 5), fixed before scoring. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on one change, a class of risk rose by 1 point (from 7 to 8) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [For a multi-cloud enterprise, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.4](/situations/multi-cloud/) - [For a company putting AI agents into production, Act ranks first at 5.6, ahead of Sonrai Security at 5.2, with Turbot and AWS Control Tower joint third at 4.2](/situations/ai-agents/) - [For a regulated company, Sonrai Security and Act share first at 5.2, with Sonrai listed first, and Turbot is third at 4.3](/situations/regulated/) ## Sources - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/aws-scale/ --- # For a multi-cloud enterprise, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.4 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. For a multi-cloud enterprise with separate teams per cloud, the top three on this situation's weights are Act Security (5.6), Sonrai Security (5.5) and AWS Control Tower (4.4). Separate teams per cloud need changes enforced through each cloud's own controls and exceptions handled in each team's tools. Act is first at 5.6, ahead of Sonrai Security at 5.5. Separate teams per cloud mean three sets of mechanisms: SCPs and RCPs in AWS, Azure Policy and deny assignments in Azure, organization policy, IAM deny and VPC Service Controls in Google Cloud. This view weights stays fixed at 15, uses the cloud's own controls at 12.5 and identity, network and AI coverage at 8.5, above their usual weights, on top of safe to enforce and class of risk at 22.5 each. The evidence differs sharply by cloud. Sonrai documents enforcement through SCPs, RCPs, Azure RBAC and Google Cloud deny policies. Act's pages name AWS and Azure boundaries, and no Google Cloud boundary mechanism was found on 2026-11-10. AWS Control Tower is AWS only. A platform that covers one cloud well can still rank below one that covers three adequately. ## The top three for this situation ### Rank 1: Act Security (5.6) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 2: Sonrai Security (5.5) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 3: AWS Control Tower (4.4) Provider guardrail service, One change, a class of risk 6 Highest score: one change, a class of risk, 6 of 10. Lowest: safe to enforce, 1. [Read the profile](/platforms/aws-control-tower/) ## The full order under these weights | Rank | Platform | Total | Removes (17.5) | Safe (22.5) | Class (22.5) | Fixed (15) | Id+Net+AI (8.5) | Native (12.5) | Audit (1.5) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.6 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.5 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.4 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Turbot](/platforms/turbot/) | 4.1 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 5 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.0 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 6 | [Apono](/platforms/apono/) | 3.7 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.3 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8 | [Wiz](/platforms/wiz/) | 3.1 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 9 | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 3.0 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 10 | [P0 Security](/platforms/p0/) | 2.8 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 11 | [Tenable Cloud Security](/platforms/tenable/) | 2.7 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 12 | [CrowdStrike](/platforms/crowdstrike/) | 2.5 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 13= | [Entitle](/platforms/entitle/) | 2.4 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 13= | [Sysdig](/platforms/sysdig/) | 2.4 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 2.2 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.5 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Published weights for this situation, the mean of this site's weights and the situation's own (removes 20, safe 15, class 15, fixed 20, id+net+ai 15, native 15, audit 0), fixed before scoring. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on safe to enforce rose by 1 point (from 5 to 6) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [For a company putting AI agents into production, Act ranks first at 5.6, ahead of Sonrai Security at 5.2, with Turbot and AWS Control Tower joint third at 4.2](/situations/ai-agents/) - [For a regulated company, Sonrai Security and Act share first at 5.2, with Sonrai listed first, and Turbot is third at 4.3](/situations/regulated/) - [For a small cloud security team, Sonrai Security and Act share first at 5.5, with Sonrai listed first, and AWS Control Tower is third at 4.4](/situations/small-team/) ## Sources - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/multi-cloud/ --- # For a company putting AI agents into production, Act ranks first at 5.6, ahead of Sonrai Security at 5.2, with Turbot and AWS Control Tower joint third at 4.2 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. For a company putting AI agents into production this year, the top three on this situation's weights are Act Security (5.6), Sonrai Security (5.2) and Turbot (4.2). An agent's reach is the sum of identity, network and AI-service access; a model that misses one misses the path. Act is first at 5.6, ahead of Sonrai Security at 5.2. An agent's reach is the sum of its identity, network and AI-service access, so this view puts identity, network and AI coverage at 18.5. Safe to enforce stays at 22.5 and class of risk at 20. The guardrails that matter here are the library's ai-services entries, 19 in all: Amazon Bedrock model allow lists and API key limits, Azure AI Services and Machine Learning network and key settings, and Vertex AI model access constraints. Act's pages describe a model that combines identity, network, resource and organization controls with endpoint and cloud AI agents, and Act says its boundary also covers the agent and the gateway or MCP server in front of it, with each change applied by the customer. SACR says Act's customer evidence for AI boundaries is less mature than for its core access control. Sonrai says it enforces least privilege for AI agent identities at the IAM layer; a network model for them: none found on 2026-11-10. ## The top three for this situation ### Rank 1: Act Security (5.6) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 2: Sonrai Security (5.2) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 3 (joint): Turbot (4.2) Guardrails as code, Evidence for auditors 6 Highest score: evidence for auditors, 6 of 10. Lowest: safe to enforce, 2. [Read the profile](/platforms/turbot/) ## The full order under these weights | Rank | Platform | Total | Removes (17.5) | Safe (22.5) | Class (20) | Fixed (10) | Id+Net+AI (18.5) | Native (7.5) | Audit (4) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1 | [Act Security](/platforms/act/) | 5.6 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 2 | [Sonrai Security](/platforms/sonrai/) | 5.2 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 3= | [Turbot](/platforms/turbot/) | 4.2 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 3= | [AWS Control Tower](/platforms/aws-control-tower/) | 4.2 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 5 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 4.0 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 6 | [Apono](/platforms/apono/) | 3.7 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Wiz](/platforms/wiz/) | 3.5 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 8 | [Britive](/platforms/britive/) | 3.3 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 9 | [Tenable Cloud Security](/platforms/tenable/) | 3.1 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 10= | [P0 Security](/platforms/p0/) | 2.8 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 10= | [CrowdStrike](/platforms/crowdstrike/) | 2.8 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 10= | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.8 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 13= | [Entitle](/platforms/entitle/) | 2.6 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 13= | [Sysdig](/platforms/sysdig/) | 2.6 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 2.5 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 2.1 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Published weights for this situation, the mean of this site's weights and the situation's own (id+net+ai 35, removes 20, safe 15, class 10, fixed 10, native 5, audit 5), fixed before scoring. ## What would change this ranking Sonrai Security would move ahead of Act Security if its score on safe to enforce rose by 2 points (from 5 to 7) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [For a regulated company, Sonrai Security and Act share first at 5.2, with Sonrai listed first, and Turbot is third at 4.3](/situations/regulated/) - [For a small cloud security team, Sonrai Security and Act share first at 5.5, with Sonrai listed first, and AWS Control Tower is third at 4.4](/situations/small-team/) - [For an AWS estate of hundreds of accounts, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.3](/situations/aws-scale/) ## Sources - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/ai-agents/ --- # For a regulated company, Sonrai Security and Act share first at 5.2, with Sonrai listed first, and Turbot is third at 4.3 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. For a regulated company that must prove access boundaries for PCI DSS, HIPAA or NIST 800-53, the top three on this situation's weights are Sonrai Security (5.2), Act Security (5.2) and Turbot (4.3). Act is joint 1st at 5.2. An auditor asks for the boundary and the continuous evidence that it held. Sonrai Security and Act share first at 5.2; Sonrai Security is listed first because it is ahead before rounding (5.22 against 5.18). What puts Sonrai ahead: removes access (5 against 3), evidence for auditors (4 against 3) and cloud's own controls (7 against 6), worth 0.56 points on these weights, against Act's lead on safe to enforce (7 against 5) and identity, network and AI (6 against 4), worth 0.52. An auditor asks for the boundary and for proof that it held. This view raises evidence for auditors to 14, stays fixed to 15 and removes to 17.5, and lowers one change for a class of risk to 20. Safe to enforce still carries 22.5, since a guardrail that breaks a regulated workload is its own finding. Sonrai and Act total 5.2 each at one decimal, so they share first place. Sonrai is listed first because it is ahead before rounding, on removes, uses the cloud's own controls and evidence for auditors. Act is ahead on safe to enforce and on identity, network and AI coverage. Sonrai has an ISO 27002 mapping page and session logs. Act names NIST 800-53, PCI DSS and HIPAA as standards whose boundaries it enforces; mapped controls or continuous evidence for them: none found on 2026-11-10. Turbot's pages state policies ready for CIS, NIST and PCI with live control state. ## The top three for this situation ### Rank 1 (joint): Sonrai Security (5.2) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 1 (joint): Act Security (5.2) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 3: Turbot (4.3) Guardrails as code, Evidence for auditors 6 Highest score: evidence for auditors, 6 of 10. Lowest: safe to enforce, 2. [Read the profile](/platforms/turbot/) ## The full order under these weights | Rank | Platform | Total | Removes (17.5) | Safe (22.5) | Class (20) | Fixed (15) | Id+Net+AI (3.5) | Native (7.5) | Audit (14) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1= | [Sonrai Security](/platforms/sonrai/) | 5.2 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 1= | [Act Security](/platforms/act/) | 5.2 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 3 | [Turbot](/platforms/turbot/) | 4.3 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 4 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.2 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 5= | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 3.7 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 5= | [Apono](/platforms/apono/) | 3.7 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Wiz](/platforms/wiz/) | 3.4 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 8 | [Britive](/platforms/britive/) | 3.2 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 9= | [Tenable Cloud Security](/platforms/tenable/) | 3.0 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 9= | [CrowdStrike](/platforms/crowdstrike/) | 3.0 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 11 | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.9 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 12 | [Sysdig](/platforms/sysdig/) | 2.8 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 13 | [P0 Security](/platforms/p0/) | 2.7 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 14= | [Entitle](/platforms/entitle/) | 2.4 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 14= | [Prisma Cloud](/platforms/prisma/) | 2.4 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.8 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Published weights for this situation, the mean of this site's weights and the situation's own (audit 25, fixed 20, removes 20, safe 15, class 10, id+net+ai 5, native 5), fixed before scoring. ## What would change this ranking Act Security would move ahead of Sonrai Security if its score on safe to enforce rose by 1 point (from 7 to 8) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [For a small cloud security team, Sonrai Security and Act share first at 5.5, with Sonrai listed first, and AWS Control Tower is third at 4.4](/situations/small-team/) - [For an AWS estate of hundreds of accounts, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.3](/situations/aws-scale/) - [For a multi-cloud enterprise, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.4](/situations/multi-cloud/) ## Sources - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/regulated/ --- # For a small cloud security team, Sonrai Security and Act share first at 5.5, with Sonrai listed first, and AWS Control Tower is third at 4.4 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. For a small cloud security team buried under posture findings it cannot clear, the top three on this situation's weights are Sonrai Security (5.5), Act Security (5.5) and AWS Control Tower (4.4). Act is joint 1st at 5.5. A team that cannot clear findings needs the product to make the change, many paths at a time, and keep it fixed. Sonrai Security and Act share first at 5.5; Sonrai Security is listed first because it is ahead before rounding (5.50 against 5.46). What puts Sonrai ahead: removes access (5 against 3), cloud's own controls (7 against 6) and evidence for auditors (4 against 3), worth 0.57 points on these weights, against Act's lead on safe to enforce (7 against 5) and identity, network and AI (6 against 4), worth 0.52. A small team cannot write and test dozens of policies by hand, so this view weights removes at 22.5 and one change for a class of risk at 27.5, with safe to enforce at 22.5. It rewards the product that carries the most work: it produces the policy, shows its effect and keeps it in place. Sonrai and Act are level at 5.5 at one decimal. Sonrai is 0.045 ahead before rounding, because its removes lead cancels Act's safe lead and its advantage on native controls and audit exceeds Act's coverage lead. AWS Control Tower is third with a catalog of more than 750 managed controls on AWS, though AWS tells users to test proactive controls themselves. ## The top three for this situation ### Rank 1 (joint): Sonrai Security (5.5) Cloud permissions firewall, One change, a class of risk 7 Highest score: one change, a class of risk, 7 of 10. Lowest: stays fixed, 4. [Read the profile](/platforms/sonrai/) ### Rank 1 (joint): Act Security (5.5) Access hardening platform, Safe to enforce 7 Highest score: safe to enforce, 7 of 10. Lowest: removes access, 3. [Read the profile](/platforms/act/) ### Rank 3: AWS Control Tower (4.4) Provider guardrail service, One change, a class of risk 6 Highest score: one change, a class of risk, 6 of 10. Lowest: safe to enforce, 1. [Read the profile](/platforms/aws-control-tower/) ## The full order under these weights | Rank | Platform | Total | Removes (22.5) | Safe (22.5) | Class (27.5) | Fixed (12.5) | Id+Net+AI (3.5) | Native (7.5) | Audit (4) | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 1= | [Sonrai Security](/platforms/sonrai/) | 5.5 | 5 | 5 | 7 | 4 | 4 | 7 | 4 | | 1= | [Act Security](/platforms/act/) | 5.5 | 3 | 7 | 7 | 4 | 6 | 6 | 3 | | 3 | [AWS Control Tower](/platforms/aws-control-tower/) | 4.4 | 5 | 1 | 6 | 5 | 4 | 6 | 4 | | 4 | [Turbot](/platforms/turbot/) | 4.2 | 5 | 2 | 5 | 5 | 5 | 3 | 6 | | 5 | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 3.9 | 4 | 6 | 3 | 2 | 4 | 5 | 2 | | 6 | [Apono](/platforms/apono/) | 3.7 | 5 | 3 | 3 | 3 | 4 | 5 | 4 | | 7 | [Britive](/platforms/britive/) | 3.2 | 5 | 1 | 3 | 3 | 4 | 5 | 4 | | 8 | [Wiz](/platforms/wiz/) | 3.0 | 3 | 4 | 1 | 4 | 6 | 2 | 6 | | 9 | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 2.8 | 3 | 4 | 1 | 4 | 2 | 4 | 2 | | 10= | [P0 Security](/platforms/p0/) | 2.7 | 5 | 0 | 2 | 3 | 4 | 5 | 3 | | 10= | [Tenable Cloud Security](/platforms/tenable/) | 2.7 | 5 | 3 | 0 | 4 | 5 | 0 | 5 | | 12= | [Entitle](/platforms/entitle/) | 2.5 | 5 | 0 | 2 | 3 | 4 | 2 | 3 | | 12= | [CrowdStrike](/platforms/crowdstrike/) | 2.5 | 3 | 4 | 0 | 4 | 4 | 0 | 6 | | 14 | [Sysdig](/platforms/sysdig/) | 2.2 | 3 | 2 | 0 | 4 | 4 | 3 | 6 | | 15 | [Prisma Cloud](/platforms/prisma/) | 2.1 | 3 | 0 | 1 | 4 | 5 | 3 | 5 | | 16 | [Microsoft](/platforms/microsoft/) | 1.4 | 2 | 0 | 1 | 2 | 6 | 0 | 5 | Published weights for this situation, the mean of this site's weights and the situation's own (removes 30, class 25, safe 15, fixed 15, native 5, id+net+ai 5, audit 5), fixed before scoring. ## What would change this ranking Act Security would move ahead of Sonrai Security if its score on one change, a class of risk rose by 1 point (from 7 to 8) with nothing else changing. A score changes only when a sourced fact changes, and every change is listed on the corrections page. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [For an AWS estate of hundreds of accounts, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.3](/situations/aws-scale/) - [For a multi-cloud enterprise, Act ranks first at 5.6, ahead of Sonrai Security at 5.5 and AWS Control Tower at 4.4](/situations/multi-cloud/) - [For a company putting AI agents into production, Act ranks first at 5.6, ahead of Sonrai Security at 5.2, with Turbot and AWS Control Tower joint third at 4.2](/situations/ai-agents/) ## Sources - sonraisecurity.com/platform: https://sonraisecurity.com/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - sonraisecurity.com/use-cases/least-privilege: https://sonraisecurity.com/use-cases/least-privilege (read 10 Nov 2026) - act.security: https://act.security (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) - softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement: https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/situations/small-team/ --- # Guides for the engineer who writes the policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Guides on writing, testing and rolling out preventive guardrails in AWS, Azure and Google Cloud, each built from provider documentation read on 10 Nov 2026, with the limits of every mechanism stated and linked to the page that states them. - [AWS SCP examples explained: what each pattern blocks and breaks](/guides/aws-scp-examples-explained/): Six jobs, 27 entries, and the breakage each one causes. - [How SCP evaluation works: allow lists, deny lists, and the limits](/guides/how-scp-evaluation-works/): Allow lists, deny lists, inheritance, exceptions and limits. - [Resource control policies: what RCPs are and how they differ from SCPs](/guides/resource-control-policies/): The control that reaches principals outside your organization. - [Permissions boundaries: capping IAM users and roles, and delegation](/guides/permissions-boundaries/): The per-identity cap, and how to delegate user creation inside it. - [Azure Policy as a guardrail: effects, enforcement mode, exemptions](/guides/azure-policy-as-a-guardrail/): Effects, DoNotEnforce, management group assignments and exemptions. - [Google Cloud organization policy: constraints, dry-run, IAM deny](/guides/google-cloud-organization-policy/): Constraints, inheritance, dry-run, custom constraints, IAM deny, VPC Service Controls. - [Preventive vs detective controls: when to block and when to report](/guides/preventive-vs-detective-controls/): Three kinds of control, what each costs, and how to choose. - [IAM Access Analyzer as a guardrail tool: before and after a change](/guides/iam-access-analyzer-as-a-guardrail-tool/): Policy validation, custom checks, unused access, and the line it does not cross. - [Testing a guardrail before you apply it: simulators and dry-run modes](/guides/testing-a-guardrail-before-you-apply-it/): What the AWS simulator, Azure DoNotEnforce and Google dry-run cover. - [Rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/): Stages, exceptions, break-glass access, and monitoring denied calls. - [AWS guardrails: Bedrock Guardrails vs account guardrails](/aws-guardrails/): Two meanings of one word, then the account kind. - [AWS Control Tower guardrails: now called controls](/aws-control-tower-guardrails/): Three control types, three implementations, three limits. - [Multi cloud governance: three hierarchies, three mechanisms](/multi-cloud-governance/): One table of the three clouds, and where they differ. - [What is cloud governance? The security case](/what-is-cloud-governance/): The security meaning, the scope of this site, and what a rule looks like. - [Cloud governance framework: six layers for security](/cloud-governance-framework/): Six layers, with the native object for each cloud. - [Guardrail exceptions and waivers in AWS, Azure and Google](/guardrail-exceptions-and-waivers/): Three exception models, one with expiry. - [Landing zone guardrails for AWS: the Control Tower baseline](/landing-zone-guardrails-aws/): Mandatory controls, the 4.0 changes, and the matching library entries. - [Azure landing zone guardrails: policy by management group](/landing-zone-guardrails-azure/): Management groups as the policy scope, per Microsoft. - [What is policy as code? Definition and where it runs](/what-is-policy-as-code/): A definition, five properties, and the four places a rule can run. - [Policy as code examples: Rego, Sentinel, Kyverno, Guard](/policy-as-code-examples/): Five real rules, attributed, with each engine's license. - [Compliance as code: rule bundles for AWS, Azure and Linux](/compliance-as-code/): Four bundle types, and which of them can block. - [Policy as code tools: five engines by the same eight fields](/policy-as-code-tools/): The engines index, alphabetical, no ranking. - [Open Policy Agent: fields, version and testing](/engines/open-policy-agent/): The OPA fields, with the limit of plan-time data. - [HashiCorp Sentinel: fields, version and testing](/engines/hashicorp-sentinel/): The Sentinel fields, with the license marked as not found. - [Cloud Custodian: fields, version and testing](/engines/cloud-custodian/): The Cloud Custodian fields, and the corrective nature of its actions. - [Kyverno: fields, version and testing](/engines/kyverno/): The Kyverno fields, and what it cannot reach. - [CloudFormation Guard: fields, version and testing](/engines/cloudformation-guard/): The Guard fields, and the Hooks route to blocking. ## Related pages - [Glossary](/glossary/) - [News](/news/) - [Preventive guardrail mechanisms in AWS, Azure and Google Cloud](/mechanisms/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/ --- # AWS SCP examples are mostly deny lists with one named exception, and each one breaks something specific Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 27 service control policy entries from AWS documentation and the AWS sample repository. They fall into six jobs, from protecting the organization to constraining Bedrock. This guide names what each pattern blocks, what it breaks in a live account, and which ones carry no exception at all. ## The 27 SCP entries sort into six jobs The AWS sample repository groups its service control policies by purpose, and the library keeps that grouping. Every entry on the [SCP mechanism page](/mechanisms/scp/) states the exact actions it denies and a documented side effect. The table below picks one entry per job. AWS documents SCPs as available only in an organization that has all features enabled, so an organization using only consolidated billing cannot use any of them. Counting the entries is a good first check on a proposed rollout. A team that wants "the standard SCPs" usually means three or four of these, not all 27, and the sections below show why the others are optional. The count is also a reminder that SCPs compete for limited space: each root, OU and account takes at most 10 attached SCPs, and each policy document is capped at 10,240 characters. | Job | Example entry | What it blocks | | --- | --- | --- | | Protect the organization | [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) | organizations:LeaveOrganization | | Contain root and IAM | [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) | Creating IAM users, access keys and login profiles | | Keep logging and detection on | [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) | Disabling detectors, changing filters and members | | Protect data and keys | [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/) | Scheduling key deletion, deleting aliases | | Limit where and what | [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/) | Every action outside the approved Regions | | Constrain AI services | [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/) | Invoking models outside the approved pattern | ## Most examples deny an action unless a privileged role is calling The common shape is a Deny statement with a condition that exempts one named role. [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) stops everyone except a privileged role or a tagged exception from creating IAM users and access keys. [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/) works the same way. AWS documents the same shape in its own example, which combines an allow of selected services with a Deny on organizations:LeaveOrganization in one policy. That shape moves the question from "who may do this" to "who owns the privileged role". The policy text carries a placeholder, [PRIVILEGED_ROLE], that the sample repository leaves for you to fill in. A policy applied with the placeholder unreplaced exempts nobody, so check the role name before attaching it to anything. Ask who holds that role, how its use is logged, and what happens when its credentials leak. ## A few examples carry no exception, so they apply to everyone Three entries have no exception condition: [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/), [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) and [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/). AWS documents that SCPs affect all users and roles in attached accounts, including the root user, so these bind the account administrator too. These are the entries to review with account owners first, because nothing in the policy gives them a way around the rule. The side effect is practical. Cleanup automation that deletes flow logs fails. A deletion scheduled with a pending window under 30 days fails. Neither shows up in a code review of the policy, which is why the library lists the breakage next to the policy and not only the intent. A fix means editing or detaching the SCP, so write down the failure mode in the change request before the attachment, not after the first ticket. ## Each pattern breaks something you can name in advance The Region deny lists approved Regions and exempts global services through a NotAction list. The README warns the example may not include every current global service, so a newly launched global service can fail until you add it. [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/) fails any upload that omits the encryption header, for every principal. Neither failure is visible from the policy name alone, and each appears only when a deployment hits it. The detection-service entries are quieter. [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) includes guardduty:ArchiveFindings and guardduty:CreateFilter, so analysts handling findings need the privileged role as well. [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/) blocks teams from adding their own Config rules. The Config entry has a further wrinkle: its README notes that AWS Control Tower already applies a similar default control, so the SCP may repeat a protection you already have. ## Treat the sample policies as drafts and test them in one OU first AWS states that it strongly recommends against attaching SCPs to the organization root without thorough testing, and suggests an OU that accounts move into one at a time. The sample repository is licensed MIT-0, so you can copy and edit the policies, and several need edits: placeholder trail names, Region lists, approved model identifiers. Plan for those edits before the review meeting, not during it. Use [how SCP evaluation works](/guides/how-scp-evaluation-works/) to predict the effect of stacking two of these across OU levels, and [testing a guardrail before you apply it](/guides/testing-a-guardrail-before-you-apply-it/) for the checks to run first. The entries themselves are listed under [the library](/library/) with the full JSON. Each entry page shows the full policy, the source file in the AWS repository and the documented side effect, so a reviewer can compare the text with the claim. ## Related pages - [How SCP evaluation works: allow lists, deny lists, and the limits](/guides/how-scp-evaluation-works/) - [Resource control policies: what RCPs are and how they differ from SCPs](/guides/resource-control-policies/) - [Permissions boundaries: capping IAM users and roles, and delegation](/guides/permissions-boundaries/) - [All of them](/guides/) ## Sources - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - AWS Organizations: SCP evaluation: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_evaluation.html (read 10 Nov 2026) - AWS Organizations: Service control policy examples: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_examples.html (read 10 Nov 2026) - AWS samples: service-control-policy-examples README: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/README.md (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/aws-scp-examples-explained/ --- # An SCP never grants access, an allow must exist at every level, and a deny at any level wins Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS documents SCP evaluation as deny by default for allow statements and deny wins at any level of the path from the organization root to the account. This guide covers allow lists, deny lists, inheritance through OUs, what SCPs never touch, and the documented size and attachment limits. ## An SCP sets a maximum and grants nothing AWS describes a service control policy as a permission guardrail that sets limits on what IAM users and roles in an organization can do. No permission is granted by an SCP. The effective permissions are the intersection of what the SCPs and resource control policies allow and what the identity-based and resource-based policies allow. Both sides must allow an action before it succeeds, and either side can stop it. A user with no identity policy has no access even when every SCP allows everything. If a permissions boundary is also present, AWS documents that the boundary, the SCP and the identity-based policy must all allow the action. See [permissions boundaries](/guides/permissions-boundaries/) for the boundary side. So a boundary adds a third condition to the same intersection. Plan each new SCP with the identity policies and boundaries it will meet in mind, not in isolation. ## Allow statements are needed at every level of the path For a permission to be allowed in an account, AWS documents that an explicit Allow must exist at every level from the root through each OU on the path to the account, including the account itself. That is why enabling SCPs attaches the AWS managed policy FullAWSAccess to every root, OU and account. Figures 1 and 2 on the AWS page show Account B losing access when the Production OU lacks the allow. If FullAWSAccess is removed at a level and not replaced, everything beneath that level is blocked. AWS says not to remove it unless you replace it with a policy that allows actions, otherwise all actions from member accounts fail. An allow list replaces FullAWSAccess with a policy naming permitted services, at the root or at each OU. AWS repeats the warning in its guidance on testing SCPs, which is worth reading before any edit to the root. ## A deny at any level cannot be overridden below it For a deny, any SCP on the path can block the permission. AWS gives the example of a deny attached to a Production OU for one service, with allows for the same service at the root and at the account: the account is still denied. Scenario 6 in the AWS page repeats the point at the root: an S3 deny at the root beats an S3 allow in a child OU. Treat a deny as fixed for the OU where it is attached, and grant exceptions through a condition inside the deny, not through an allow lower down. This is why deny lists are the easier pattern to reason about. Each deny holds wherever it is attached, and a deny-only SCP at the root with no allow statement blocks every service, so keep FullAWSAccess in place when you add denies. AWS also notes that relying only on allow statements can lead to unintended access, because broader or overlapping allows can override narrower ones. Writing denies as narrow conditions inside one statement keeps that review short. ## Some principals and tasks are outside SCP reach SCPs do not affect users or roles in the management account, and they do not affect any service-linked role. They apply to member accounts that are designated as delegated administrators. They also do not affect resource-based policies directly or principals from accounts outside the organization: a bucket policy that grants an outside account is not limited by the bucket owner account SCP. A control that must hold for the management account's own users needs a different mechanism. AWS lists further tasks SCPs cannot restrict, such as registering for the Enterprise support plan as the root user and providing trusted signer functionality for CloudFront private content. The gap for outside principals is what [resource control policies](/guides/resource-control-policies/) were added to cover. Those exclusions explain why each library entry says what it does not cover. Read the section on tasks and entities not restricted by SCPs before you promise coverage to an auditor. ## The documented limits shape how you layer policies An SCP document can be up to 10,240 characters, and the sizes count white space. Each root, OU and account can have at most 10 SCPs attached directly, every entity must keep at least one, and OUs can nest five levels deep under a root. AWS documents these as hard limits. AWS states that all policy limits are hard limits, so there is no quota request to file. Inherited policies do not count against the attachment limit, so many small policies near the top and few at the bottom stay within it. If you disable the SCP policy type in a root, every SCP is detached from every entity, and the attachments are lost and not automatically recoverable. Keep the policy JSON in version control so the state can be rebuilt. ## Related pages - [Resource control policies: what RCPs are and how they differ from SCPs](/guides/resource-control-policies/) - [Permissions boundaries: capping IAM users and roles, and delegation](/guides/permissions-boundaries/) - [Azure Policy as a guardrail: effects, enforcement mode, exemptions](/guides/azure-policy-as-a-guardrail/) - [All of them](/guides/) ## Sources - AWS Organizations: SCP evaluation: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_evaluation.html (read 10 Nov 2026) - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - AWS Organizations: Quotas and service limits: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - IAM: Permissions boundaries: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/how-scp-evaluation-works/ --- # RCPs cap what anyone can do to your resources, including callers outside your organization Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS added resource control policies to set the maximum permissions on resources in member accounts, whoever the caller is. SCPs limit your own principals; RCPs limit access to your resources, including from outside the organization. This guide covers scope, the supported services, the limits and the data perimeter use. ## RCPs limit resources, SCPs limit principals AWS gives a one-line rule. Use an SCP to limit the permissions of IAM principals inside your member accounts. Use an RCP to restrict principals outside your organization that request access to resources in your member accounts. Both are organization policies and neither grants permissions. SCPs sit on the identity side of a request and RCPs on the resource side, so a request to read an S3 object meets the SCPs of the caller's account, when it is in an organization, and the RCPs of the account that owns the bucket. An RCP is closest in kind to a resource-based policy, but it is attached to the account, OU or root and sets a ceiling over the resource-based policies below it. The effective permissions are the intersection of the RCPs and SCPs with the identity-based and resource-based policies. See [the RCP mechanism page](/mechanisms/rcp/) for the library entries. AWS states the job of RCPs in one sentence: central control over the maximum available permissions for resources. ## RCPs apply whoever the caller is AWS documents that RCPs affect the effective permissions of principals reaching a resource in a member account whether or not they belong to the organization, and that includes root users. An RCP on a bucket in Account A applies when a user from outside the organization reads it. It does not apply to resources in the outside account when your users reach those. That reach is the reason the library flags breakage for partners. [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/) denies requests over an unencrypted connection, including from trusted partners. [Require TLS 1.2 or later for S3](/guardrails/aws-rcp-s3-minimum-tls-1-2/) denies older clients wherever they sit. Test a candidate RCP against the partners and vendors that call your buckets and queues, not only against your own workloads. Their failures show up as Access Denied errors in your CloudTrail logs. ## What RCPs cannot restrict, and where they apply RCPs apply to a subset of AWS services. The AWS list includes Amazon S3, AWS STS, AWS KMS, Amazon SQS, AWS Secrets Manager, Amazon DynamoDB, Amazon ECR, Amazon Cognito and others, and a service missing from the list is outside the policy. A service that is not on the list is not covered, so a control for it needs a resource-based policy or a different mechanism. Check the current list on the AWS page at each review. AWS also documents that RCPs do not affect resources in the management account, calls made by service-linked roles, AWS managed keys in KMS, or the kms:RetireGrant permission. Like SCPs, they cover member accounts designated as delegated administrators. A control that must cover the management account needs another mechanism. Management account resources are the one gap to design around: protect them with resource-based policies and tight access to the account itself. ## The limits are tighter than for SCPs An RCP can be up to 5,120 characters, half the SCP maximum. Only five can be attached to the root, to an OU or to an account, and the AWS managed RCPFullAWSAccess policy is attached automatically when you enable RCPs. You cannot detach it, and it counts toward the five. These counts are why one policy per control fills the slots quickly. In practice you get four of your own at each level. The library entries are small single-purpose policies, such as [Deny S3 customer-provided encryption keys](/guardrails/aws-rcp-deny-s3-sse-c/), that you can merge when you run short. AWS documents that all of these limits are hard limits. Put the rules that never change near the root and the exceptions near the accounts that need them, so inherited policies carry most of the weight without counting against the attachment limit. ## RCPs are the enforcement half of a data perimeter The AWS whitepaper describes a data perimeter as a set of always-on guardrails around identities, resources and networks. The RCP sample README states the aim as ensuring only trusted identities access trusted resources from expected networks, and it carries an identity perimeter RCP for the first objective. Together the two policy types give a perimeter its pair: trusted identities on one side and trusted resources on the other. The same pairing appears across the library: SCPs constrain your principals, RCPs constrain access to your resources. For rollout, AWS suggests attaching RCPs to individual test accounts first, then moving up through OUs, and reviewing CloudTrail for Access Denied errors. [Testing a guardrail first](/guides/testing-a-guardrail-before-you-apply-it/) covers why the policy simulator is no help here. Run the first test in a sandbox OU, then move to production OUs while watching for Access Denied errors from callers you did not know about. ## Related pages - [Permissions boundaries: capping IAM users and roles, and delegation](/guides/permissions-boundaries/) - [Azure Policy as a guardrail: effects, enforcement mode, exemptions](/guides/azure-policy-as-a-guardrail/) - [Google Cloud organization policy: constraints, dry-run, IAM deny](/guides/google-cloud-organization-policy/) - [All of them](/guides/) ## Sources - AWS Organizations: Resource control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - AWS Organizations: Quotas and service limits: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - AWS whitepaper: Building a Data Perimeter on AWS: https://docs.aws.amazon.com/whitepapers/latest/building-a-data-perimeter-on-aws/building-a-data-perimeter-on-aws.html (read 10 Nov 2026) - AWS Organizations: Resource control policy examples: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/resource-control-policies/ --- # A permissions boundary caps what an identity policy can grant and makes delegation safe to hand out Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS documents a permissions boundary as a managed policy that sets the maximum permissions an identity-based policy can grant to one IAM user or role. This guide covers how boundaries combine with identity policies and SCPs, the resource-based policy exceptions, and the delegated user creation pattern in the library. ## A boundary caps one user or role and grants nothing A permissions boundary is a managed policy, AWS managed or customer managed, set on one IAM user or role. The entity can perform only the actions allowed by both its identity-based policies and its boundary. The boundary alone grants nothing. The managed policy can be an AWS managed policy or a customer managed policy. It is set per entity, so each user or role that needs the cap must carry it, and an identity without a boundary has none. AWS gives a user named Shirley a boundary that allows s3:*, cloudwatch:* and ec2:*. A separate policy allowing iam:CreateUser fails, because the boundary does not allow the action. The library keeps this as [Permissions boundary limiting a user to S3, CloudWatch and EC2](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/). AWS adds that, given those two policies, Shirley has no permission to perform any operation in AWS until a permissions policy allows actions in other services. ## Boundaries, SCPs and identity policies must all agree Within an account, AWS lists identity-based policies, resource-based policies, permissions boundaries, Organizations SCPs and session policies as the policy types that affect effective permissions. If any one of them explicitly denies an operation, the request is denied. When a boundary and an SCP are both present, the boundary, the SCP and the identity-based policy must all allow the action. Check all of those layers before you decide that an access failure is a missing grant. The scopes differ. An SCP caps every principal in an account or OU, which suits a baseline. A boundary caps one identity, which suits a person or role that is allowed to create other identities. For the organization-wide rules, see [how SCP evaluation works](/guides/how-scp-evaluation-works/). A tight boundary on a role that already sits under an SCP is therefore redundant for the actions the SCP blocks, but it still protects the identity if the role moves to an account with a looser SCP. ## Resource-based policies can bypass a boundary in documented cases Within the same account, a resource-based policy that grants permissions to an IAM user ARN is not limited by an implicit deny in the user identity-based policy or boundary. A grant to a role session ARN, or to an AWS STS federated user session, goes directly to the session and is not limited by the boundary either. A grant to the role ARN itself is limited by the boundary. AWS also warns against a NotPrincipal element with a Deny effect on resources used by identities that have a boundary: it denies every principal that has a boundary attached. The recommendation is the ArnNotEquals condition with aws:PrincipalArn. Rewrite any such statement before you add boundaries to the identities it covers; the failure otherwise shows up as lost access for principals that looked allowed in review. ## Delegation is the strongest use AWS describes a boundary as the way to let an IAM user create other users inside limits. In its example, an administrator lets a colleague create users, on the condition that those users cannot manage IAM, cannot reach a logs bucket and an instance, and cannot remove their own boundary. The boundary therefore does the delegating work that an SCP cannot do for a single person. The library entry is [Permissions boundary for delegated IAM user creation](/guardrails/aws-pb-delegated-user-creation-boundary/). It allows creating or changing users only when the XCompanyBoundaries boundary is applied, and denies editing the boundary policies or deleting a user boundary. AWS notes the delegate can create a user who can reach S3 resources the delegate cannot, so delegation trusts the delegate with that access. Review that trust at least as often as the policy itself. ## Where a boundary fits next to the other mechanisms A boundary attaches to one user or role, so it fits identities that create other identities. It does not protect a resource from outside callers, which is the [resource control policy](/guides/resource-control-policies/) job, and it does not set a baseline for a whole account, which is the SCP job. Which one to reach for depends on whether the risk is an identity creating identities or a resource reached from outside. The IAM policy simulator can test one permissions boundary at a time. A boundary change takes effect on every identity that carries it, so simulate the new version before you attach it. [The testing guide](/guides/testing-a-guardrail-before-you-apply-it/) covers how, and the [permissions boundary mechanism page](/mechanisms/permissions-boundary/) lists both library entries. Keep the boundary policy in version control, because a change to the managed policy applies to every identity that carries it. ## Related pages - [Azure Policy as a guardrail: effects, enforcement mode, exemptions](/guides/azure-policy-as-a-guardrail/) - [Google Cloud organization policy: constraints, dry-run, IAM deny](/guides/google-cloud-organization-policy/) - [Preventive vs detective controls: when to block and when to report](/guides/preventive-vs-detective-controls/) - [All of them](/guides/) ## Sources - IAM: Permissions boundaries for IAM entities: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - IAM: Policy testing with the IAM policy simulator: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) - AWS Organizations: SCP evaluation: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_evaluation.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/permissions-boundaries/ --- # Azure Policy blocks only when the effect is deny, so effect, enforcement mode and scope decide what a guardrail does Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure Policy evaluates resources against definitions and acts through an effect: deny, audit, modify or deployIfNotExists. Enforcement mode, assignment scope at management groups and exemptions then decide where and when the effect applies. This guide covers each, with the documented behavior and the traps. ## Each definition has one effect, and only some effects block Each Azure Policy definition has a single effect in its policy rule, and the effect decides what happens on a match. The deny effect prevents a resource request that does not match standards: for a Resource Manager mode, the request is returned as a 403 Forbidden before it reaches the resource provider. For existing resources, a match is marked non-compliant. The audit effect creates a warning event in the activity log and marks the resource non-compliant, but it does not stop the request. Library entries such as [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/) take the effect as a parameter, so the same definition can audit first and deny later. Switching a definition from audit to deny is then a parameter change at assignment, with no rewrite of the rule. ## Modify and deployIfNotExists fix a resource instead of refusing it The modify effect adds, updates or removes properties or tags on a resource during creation or update, and a remediation task can fix existing resources. A deployIfNotExists definition runs a template deployment after the resource provider returns success, when the related resource is missing. Assignments with either effect require a managed identity to remediate. Microsoft also notes that PATCH requests that only modify tags limit evaluation to policies whose conditions inspect tags. Evaluation order matters. Azure checks disabled first, then append and modify, then deny, then audit. Because modify runs before deny, a change it makes can prevent a deny or audit from triggering. [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/) is a library example of the deploy pattern. Grant that managed identity the permissions the template or the change requires before you assign the policy, not after the first failed remediation. ## Enforcement mode lets you assign a policy without enforcing it The enforcementMode property controls the outcome on resources without initiating the effect or writing activity log entries. The default value is Default, which enforces on create and update. DoNotEnforce evaluates the assignment and reports compliance but does not enforce the effect. The mode applies to the assignment, not the definition, so one definition can be enforced in production and unenforced in a test subscription at the same time. Microsoft documents DoNotEnforce as useful for testing a new assignment, reviewing compliance impact or validating logic before enforcement. Remediation tasks for deployIfNotExists can still be started in DoNotEnforce. A third value, Enroll, enforces only on resources enrolled in the assignment, which suits adding subscopes in batches. When no enrollment resource exists for a scope, evaluation behaves as if the assignment were DoNotEnforce by default, though Microsoft says that behavior is configurable. ## Assign at a management group and let it inherit Management groups are a governance scope above subscriptions, and conditions applied to a management group cascade by inheritance to all of its subscriptions. A single directory supports 10,000 management groups and a tree up to six levels deep. See [the management groups mechanism](/mechanisms/management-groups/) and the library example [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/). That inheritance is the point of the mechanism: one assignment replaces many. Set the assignment at the lowest group that covers your pilot subscriptions first. Effects layer by most restrictive. Microsoft gives an example of a deny at subscription level and an audit at resource group level: each assignment is evaluated on its own, and a resource cannot slip through a gap between scopes. Several assignments with a deny can block a resource that no single one of them seems to cover. Review every assignment that touches a scope before you conclude that a deny is missing. ## Exclusions and exemptions are different tools An exclusion, set with notScopes on the assignment, removes a scope from evaluation and from the compliance count. Microsoft suggests exclusions for broad, permanent bypass such as a test environment. An exemption exempts a resource hierarchy or one resource and keeps it tracked. The two are not interchangeable, and the choice changes what compliance reports show. Record the reason for each exclusion in the assignment description. An exemption has a category, Mitigated or Waiver, and an optional expiresOn date. When the date passes, the exemption object is kept for records but is no longer honored. For production, prefer a dated Waiver with a description. The [policy exemption](/glossary/policy-exemption/) and [enforcement mode](/glossary/enforcement-mode/) glossary entries have the exact terms. The waiver example in Microsoft's documentation describes a resource planned for deletion by the end of the quarter. ## Related pages - [Google Cloud organization policy: constraints, dry-run, IAM deny](/guides/google-cloud-organization-policy/) - [Preventive vs detective controls: when to block and when to report](/guides/preventive-vs-detective-controls/) - [IAM Access Analyzer as a guardrail tool: before and after a change](/guides/iam-access-analyzer-as-a-guardrail-tool/) - [All of them](/guides/) ## Sources - Azure Policy: Definition effects basics: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-basics (read 10 Nov 2026) - Azure Policy: Deny effect: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - Azure Policy: Audit effect: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-audit (read 10 Nov 2026) - Azure Policy: Modify effect: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - Azure Policy: deployIfNotExists effect: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deploy-if-not-exists (read 10 Nov 2026) - Azure Policy: Assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) - Azure Policy: Exemption structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/exemption-structure (read 10 Nov 2026) - Azure management groups overview: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/azure-policy-as-a-guardrail/ --- # Google Cloud splits guardrails into four layers: organization policy for what, IAM deny for who, and perimeters for data Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud organization policy restricts how resources can be configured through constraints set on an organization, folder or project. This guide covers constraint types, inheritance and merging, dry-run mode, custom constraints, and how IAM deny policies and VPC Service Controls cover the cases that organization policy does not. ## Organization policy controls what, IAM controls who Google describes the split plainly: IAM focuses on who, and lets an administrator say which principals can act on which resources. Organization Policy focuses on what, and sets restrictions on how resources can be configured. A policy sets one constraint on an organization, folder or project and its descendants inherit it. Both are inherited down the hierarchy, so a setting on the organization reaches every project. A constraint is the blueprint of a controlled behavior, and the organization policy is the rule set that enforces it. The library groups the entries under [the organization policy mechanism](/mechanisms/org-policy/). A policy can hold several rules, for example one that enforces a constraint only on resources tagged environment=development. The constraint list is long, and Google frames it as fine-grained control over specific resource behaviors. ## Constraints come as managed, legacy managed and custom Managed constraints are newer constraints with list or boolean parameters. Legacy managed constraints use list rules, a collection of allowed or denied values, or boolean rules, which can allow all values, deny all values or check enforcement. Google notes managed constraints are designed to replace equivalent legacy ones. Managed constraints support simulation in Policy Simulator by default, and the documentation notes any that do not. Custom constraints are defined by your organization in a YAML file that names the resources, methods, conditions and actions. A custom constraint ID must start with custom. and cannot change after creation. [The custom organization policy mechanism](/mechanisms/custom-org-policy/) holds the library example that limits IAM role grants to organization members. Custom constraints exist only for the services Google lists as supporting them, so check the supported-service list before you plan one. ## Inheritance merges or replaces, and deny wins By default descendants inherit the policy of an ancestor. A child can set inheritFromParent to true, in which case the parent policy is merged with its own, or to false, in which case it ignores the parent and uses the constraint default unless it sets its own values. Google states that the default behavior is never merged. Read the effective policy at the project, not the policy at the organization. When values are merged and conflict, the DENY value takes precedence. Managed constraints inherited from a parent are not merged: the resource enforces the constraint as set on the parent or overrides it with its own policy. If a constraint has a Google-managed default that restricts an operation, the operation is restricted even if you never wrote a policy for it. ## Dry-run mode logs violations without denying them An organization policy in dry-run mode is created and enforced like any other, but violations are audit logged and the violating actions are not denied. You need billing enabled on the project, the Organization policy administrator role and the Logs Configuration Writer role. Only custom constraints, managed constraints and certain legacy constraints support dry-run. Dry-run rules are set apart from the live policy on the constraint. Policy Simulator for Organization Policy goes one step earlier. It previews a new custom or managed constraint and lists the resources that violate the proposed policy before it is enforced, so you can reconfigure resources, request exceptions or narrow the scope. Run it first, then dry-run, then enforce. If a constraint is not available in dry-run, the simulator or a lower environment is the only preview. ## IAM deny and VPC Service Controls cover what constraints do not IAM deny policies attach to a project, folder or organization and hold deny rules for principals and permissions. Google documents that IAM always checks relevant deny policies before allow policies, so a denied permission cannot be restored by granting a role. The library has [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/) as an example. Deny policies are eventually consistent, so allow time for a change to propagate. VPC Service Controls is a separate layer around services: a service perimeter controls data egress across a boundary, independent of IAM, and Google recommends using both. It has its own dry run mode in which requests that violate the perimeter are logged and not denied. See [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/). The library lists both under their own mechanism pages. ## Principal access boundaries limit the resources a principal can reach Google documents another control that sits beside deny: principal access boundary policies. By default, principals are eligible to access any Google Cloud resource. A Principal Access Boundary policy defines the resources a principal is eligible to access, and if a principal is ineligible, access is limited regardless of the roles it holds. Google lists this as part of IAM, not of organization policy, which is why it appears here as a separate layer. These policies attach to principal sets, unlike allow policies, which attach to resources. Google gives preventing principals from reaching resources in other organizations as an example, which can help prevent phishing. See [Principal access boundary limiting the resources principals can access](/guardrails/gcp-pab-eligible-resources/) and [the mechanism page](/mechanisms/principal-access-boundary/). Check whether the principal sets you need are supported before you design around it. Both are small checks. ## Related pages - [Preventive vs detective controls: when to block and when to report](/guides/preventive-vs-detective-controls/) - [IAM Access Analyzer as a guardrail tool: before and after a change](/guides/iam-access-analyzer-as-a-guardrail-tool/) - [Testing a guardrail before you apply it: simulators and dry-run modes](/guides/testing-a-guardrail-before-you-apply-it/) - [All of them](/guides/) ## Sources - Google Cloud: Introduction to the Organization Policy Service: https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - Google Cloud: Understanding hierarchy evaluation: https://cloud.google.com/resource-manager/docs/organization-policy/understanding-hierarchy (read 10 Nov 2026) - Google Cloud: Test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) - Google Cloud: Test organization policy changes with Policy Simulator: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) - Google Cloud: Create custom constraints: https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints (read 10 Nov 2026) - Google Cloud: IAM deny policies: https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) - Google Cloud: VPC Service Controls overview: https://cloud.google.com/vpc-service-controls/docs/overview (read 10 Nov 2026) - Google Cloud: Dry run mode for service perimeters: https://cloud.google.com/vpc-service-controls/docs/dry-run-mode (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/google-cloud-organization-policy/ --- # Block what cannot be undone, report what can, and use a report-only phase to learn which is which Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A preventive control disallows the action; a detective control finds the noncompliant resource afterward; a proactive control checks a resource before deployment. AWS Control Tower, Azure Policy and Google Cloud each document all three. This guide sets out the differences, the cost of each, and a rule for choosing. ## Preventive controls stop the request, detective controls report afterward AWS Control Tower defines a preventive control as one that keeps accounts compliant because it disallows actions that lead to policy violations. Its status is enforced or not enabled, and it is implemented with service control policies or resource control policies. A detective control finds noncompliance of resources and alerts through a dashboard; its status is clear, in violation or not enabled. The practical difference is timing. A preventive control fails a request and the engineer sees an error. A detective control lets the change land and surfaces it later, and the resource exists in the meantime. The [preventive control](/glossary/preventive-control/) and [detective control](/glossary/detective-control/) glossary entries have the exact terms. Each has a cost that the other avoids, and a mature program runs both on the same rule. ## A proactive control checks the template before anything is created AWS Control Tower documents a third kind, the proactive control, which checks resources before they are deployed. It is implemented as CloudFormation hooks that run on preCreate and preUpdate. AWS notes proactive controls may not affect requests made directly through the console, the API, the SDKs or other infrastructure-as-code tools. AWS sorts proactive controls into four categories, and the reference guide groups them by AWS service. That scope limit is why a proactive control is not a substitute for a preventive one. It catches mistakes early in a pipeline that uses CloudFormation, and it leaves every other path open. Pair it with an SCP or RCP when the rule must hold on every path, and with a detective control when you need to see what already exists. ## The same split exists in Azure and Google Cloud Azure Policy expresses it as an effect. Deny returns a 403 before the request reaches the resource provider. Audit creates a warning in the activity log and marks the resource non-compliant without stopping the request. The enforcement mode DoNotEnforce goes further, and evaluates the assignment without triggering the effect. Microsoft also documents that auditIfNotExists and deployIfNotExists are often interchangeable, which gives a path from detecting to correcting. Google Cloud uses dry-run mode: an organization policy in dry-run is enforced like any other, but violations are audit logged and the violating actions are not denied. VPC Service Controls offers the same for perimeters. In all three clouds, the detective setting is the first step toward the preventive one. Dry-run is available only for the constraint types Google lists, so check support first. ## Choose by what an undo costs Block when the action cannot be undone or the damage lands before anyone can react: leaving the organization, scheduling KMS key deletion, disabling logging, making a snapshot public, running a workload in a Region you have not approved. The library marks these in the [SCP mechanism](/mechanisms/scp/) and [declarative policy](/mechanisms/declarative-policy/) pages. In each case the harm is done before a report could arrive. Report when the state is fixable and the rule is still being learned: tag conventions, naming, a setting that varies by team. A detective control costs a finding to triage, and a preventive one costs a failed deployment and a support ticket. A rule that has not run in report mode for one cycle is a risk to ship as a block. ## Run every new rule detective first, then promote it The promotion path is documented by all three providers: AWS recommends testing an SCP in an OU before the root, Azure documents DoNotEnforce for validating logic before enforcement, and Google documents dry-run to monitor how a policy change affects workflows before it is enforced. Each of the three is a staging step, and none is a permanent state. Skipping the step is the usual cause of a rollback. Keep the detective control after promotion. A deny stops the paths it covers, and a detective control finds what slipped past through the management account, a service-linked role or an exempted scope. The [rollout guide](/guides/rolling-out-guardrails-without-breaking-production/) turns this into stages. Reviewing detective findings monthly also catches drift in the exception list, since each exempted account or role is a place the block does not reach. ## The terms line up across the three clouds The table maps the documented names. Providers use different words for the same three roles, and a rule is easiest to move between clouds when you map it by role first. Use the row for your cloud when you plan the promotion path. The Azure column has one gap: a pre-deployment check was not found in the pages read on 2026-11-10, so an Azure pipeline check has to be added by other means. | Cloud | Preventive | Detective or report-only | Before deployment | | --- | --- | --- | --- | | AWS | SCP, RCP, declarative policy | Control Tower detective controls on AWS Config rules | Proactive controls as CloudFormation hooks | | Azure | Policy deny effect, deny assignment, lock | Policy audit effect, DoNotEnforce mode | Not found in the pages read | | Google Cloud | Organization policy, IAM deny, perimeter | Dry-run mode, perimeter dry run | Policy Simulator | ## Related pages - [IAM Access Analyzer as a guardrail tool: before and after a change](/guides/iam-access-analyzer-as-a-guardrail-tool/) - [Testing a guardrail before you apply it: simulators and dry-run modes](/guides/testing-a-guardrail-before-you-apply-it/) - [Rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/) - [All of them](/guides/) ## Sources - AWS Control Tower: Preventive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html (read 10 Nov 2026) - AWS Control Tower: Detective controls: https://docs.aws.amazon.com/controltower/latest/controlreference/detective-controls.html (read 10 Nov 2026) - AWS Control Tower: Proactive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) - Azure Policy: Definition effects basics: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-basics (read 10 Nov 2026) - Azure Policy: Assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) - Google Cloud: Test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/preventive-vs-detective-controls/ --- # IAM Access Analyzer checks a guardrail policy before you apply it and finds access it missed afterward, and leaves enforcement to SCPs, RCPs and boundaries Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. IAM Access Analyzer validates policies against grammar and AWS best practices, runs custom checks for new access, and keeps analyzers for external, internal and unused access. This guide separates what it does before a policy change from what it reports afterward, and what it leaves to SCPs and RCPs. ## Policy validation is the first check on any guardrail JSON Policy validation checks a policy against IAM policy grammar and AWS best practices. Findings come as security warnings, errors, general warnings and suggestions. In the IAM console they appear for managed policies and for inline user and role policies, and AWS notes they are not generated for inline group policies. Run it on every guardrail file, since a syntax error found at attach time is slower to fix. The check reference includes checks written for SCPs. One example is the error "SCP syntax error action wildcard": SCP actions can include wildcards only at the end of a string. A guardrail JSON file copied from the library can be run through validation before it is attached, and catching a bad wildcard here is cheaper than a failed attachment. Each finding also names the location of the reported issue. ## Custom policy checks answer whether an edit grants new access Custom policy checks test a policy against your own security standard. In the console, you can edit a policy and check whether the new version grants new access compared with the existing one. Through the CLI and API, check-no-new-access does the same comparison, check-access-not-granted verifies that named actions are not allowed, and check-no-public-access tests whether a resource policy can grant public access. AWS notes that a charge applies to each check for new access. For a team that edits guardrail policies through a pipeline, check-access-not-granted is the closest fit: it asserts that critical actions stay unreachable after every change. Put it in the same pipeline step that validates the JSON, and fail the build when a check returns a finding, so an IAM policy edit cannot widen access by accident. ## External and internal access analyzers preview resource policies An external access analyzer uses logic-based reasoning on resource-based policies and produces a finding for each resource shared outside your account or organization. AWS says the findings can be used to preview how a policy affects public and cross-account access before deploying resource permissions. External analyzers analyze only policies applied to resources in the same Region, so create one in each Region where you use resources. That preview is the closest AWS tool for the situation an RCP addresses, access to a resource from outside. Internal access analyzers cover access from principals within your own organization to selected resources, such as S3 buckets and DynamoDB tables. Neither type changes any policy. They report. Internal access findings show access paths between principals and resources, which helps verify that your access controls work as intended. ## Unused access findings show what a guardrail should cover next An unused access analyzer continuously monitors IAM roles and users in your organization and produces findings for unused roles, unused access keys and unused passwords. For active roles and users, the findings show unused services and actions. Findings are organized in a dashboard that highlights the accounts and resources with the most findings and breaks them down by type, which gives a review meeting a ranked list to work from. Those findings feed the next guardrail. A service no role has used in the review period is a candidate for an SCP deny, and the SCP page makes a related point: service last accessed data shows which services an SCP allows but nobody uses. Policy generation can also build an IAM policy from CloudTrail activity over a date range, which you then attach to refine one identity. ## What it leaves to other mechanisms IAM Access Analyzer reports and validates. AWS does not describe it in the request path, so it is no substitute for [an SCP](/mechanisms/scp/), [an RCP](/mechanisms/rcp/) or [a permissions boundary](/mechanisms/permissions-boundary/). Findings arrive after access exists, which makes it a detective tool, in the sense used in [preventive vs detective controls](/guides/preventive-vs-detective-controls/). Treat a finding as a prompt to open a change, not as a control. Blocking is left to those mechanisms. Use it at two points: before an edit, to validate syntax and prove no new access was granted, and on a schedule afterward, to find exposure and unused permissions. The policy simulator, covered in [testing a guardrail first](/guides/testing-a-guardrail-before-you-apply-it/), answers a different question: whether a specific request is allowed. Schedule the second use monthly, and attach an owner's name to each class of finding. ## Cost and scope are documented and worth checking first AWS states that a charge applies to each custom check for new access, and that unused access analysis is charged by the number of IAM roles and users analyzed, with a charge for each unused access analyzer you create. Read the IAM Access Analyzer pricing page before you add checks to a pipeline that runs on every commit. Budget both. For unused access, AWS says findings do not change by Region, so you do not need an analyzer in each Region. External access analyzers are per Region. Plan the analyzers by account and Region before you create them, and decide who reads the findings. Decide who reads the findings. Assign a named owner for each analyzer type. ## Related pages - [Testing a guardrail before you apply it: simulators and dry-run modes](/guides/testing-a-guardrail-before-you-apply-it/) - [Rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/) - [AWS guardrails: Bedrock Guardrails vs account guardrails](/aws-guardrails/) - [All of them](/guides/) ## Sources - IAM: Using IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 10 Nov 2026) - IAM: Validate policies with IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-policy-validation.html (read 10 Nov 2026) - IAM: Custom policy checks: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-custom-policy-checks.html (read 10 Nov 2026) - IAM: Access Analyzer policy check reference: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-reference-policy-checks.html (read 10 Nov 2026) - IAM: Generate policies based on access activity: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-policy-generation.html (read 10 Nov 2026) - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/iam-access-analyzer-as-a-guardrail-tool/ --- # Each cloud gives you a way to test a guardrail first, and each has a documented gap you must cover by hand Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The IAM policy simulator evaluates SCPs and permissions boundaries but not RCPs; Azure Policy offers DoNotEnforce; Google Cloud offers dry-run mode and Policy Simulator for organization policy. This guide sets out what each tester covers, the documented gaps, and the order to run them in. ## The IAM policy simulator covers SCPs and boundaries, not RCPs The IAM policy simulator evaluates identity-based policies, permissions boundaries, SCPs and resource-based policies that you provide, without sending a real request. Each simulation returns allowed or denied per action and resource. You can simulate one permissions boundary at a time, and SCP testing is available when the account belongs to an organization. Cross-account simulations return per-policy-type decisions. Custom mode lets you paste a draft policy without saving it to the account. AWS states that the simulator does not support resource control policies. A new RCP therefore has no simulator path, and the documented alternative is to attach it to a test account and review CloudTrail for Access Denied errors. The simulator also tells you whether an SCP allows or denies an action but, for security reasons, does not surface the matched statements. ## Simulator results can differ from the live environment AWS warns that results can differ from live behavior and recommends checking your policies against the live environment afterward. Documented differences include VPC endpoint policies, role chaining and multiple resource-based policies on one resource. The simulator uses no real context values from production. Treat a simulator pass as evidence of syntax and intent, and the live check as the real test. It fills in principal and organization context keys itself, such as aws:PrincipalAccount and aws:PrincipalOrgID, so SCPs with those conditions evaluate correctly. You supply the rest. In the console, a condition key referenced only by an SCP cannot be set, so use the CLI or API for a Region-deny SCP that depends on aws:RequestedRegion. Supply a value for each key your policies reference. ## Azure DoNotEnforce tests the assignment on real resources Azure Policy takes a different route: assign the real policy in DoNotEnforce mode. The effect is not enforced on create or update and no activity log entry is written for it, but evaluation still produces compliance results, so you see which existing resources the policy would have flagged. Nothing in the mode changes the resources themselves. Do not rely on that alone. Microsoft documents DoNotEnforce as the way to evaluate an assignment before enforcement. Remediation tasks for deployIfNotExists can still run in this mode, so check whether the assignment includes one. Read compliance results for a full evaluation cycle before switching to the default mode, because the standard compliance evaluation cycle runs once every 24 hours. Moving from DoNotEnforce to Default is a one-property change to the assignment. ## Google offers two tests for organization policy and one for perimeters Policy Simulator for Organization Policy previews a new custom or managed constraint and lists the resources that would violate it, before the policy is enforced. Dry-run mode then enforces a policy as audit only: violations are logged and the violating actions are not denied. Dry-run needs billing enabled on the project. Run the simulator on the proposed constraint and read the list of violating resources. Dry-run supports custom constraints, managed constraints and certain legacy constraints, so some older constraints cannot be trialled this way. VPC Service Controls has its own dry run mode for perimeters, where violating requests are logged and not denied. The deny-policy and principal access boundary testing pages were not available when read on 2026-11-10, so those need a lower-environment trial. Record which tests were not possible. ## Run them in a fixed order First validate syntax with [IAM Access Analyzer](/guides/iam-access-analyzer-as-a-guardrail-tool/). Then simulate, or run the cloud native preview. Then enforce in report-only mode where one exists, and read the logs for one full cycle. Then apply to one account, subscription or project, and only then to its parent. Each step narrows the blast radius if the test missed something. Skip none of them for a policy that touches production data. Write the expected result before each run. A simulation that agrees with a prediction is evidence, and a result you did not predict is the finding. [The rollout guide](/guides/rolling-out-guardrails-without-breaking-production/) places these tests inside a staged rollout. Keep the written predictions with the change record, so the approver can see what was expected, what was observed and what was left untested in a lower environment. ## What each tester can and cannot show The table below sets the three clouds side by side. Read the last column as the list of things you still have to check by another route before you widen a guardrail to production. Fill in the last column for your own estate, because an untested path is where the next incident starts, and write the manual check next to each gap before you ask for approval. | Cloud | Preview tool | Report-only mode | Documented gap | | --- | --- | --- | --- | | AWS | IAM policy simulator, for SCPs and boundaries | None found on 2026-11-10 for SCPs | No RCP support; SCP matched statements not shown | | Azure | Compliance results from DoNotEnforce | DoNotEnforce, Enroll | Remediation tasks can still run | | Google Cloud | Policy Simulator for Organization Policy | Dry-run mode, perimeter dry run | Only some constraint types support dry-run | ## Related pages - [Rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/) - [AWS guardrails: Bedrock Guardrails vs account guardrails](/aws-guardrails/) - [AWS Control Tower guardrails: now called controls](/aws-control-tower-guardrails/) - [All of them](/guides/) ## Sources - IAM: Policy testing with the IAM policy simulator: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) - AWS Organizations: Resource control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - Azure Policy: Assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) - Azure Policy: Overview: https://learn.microsoft.com/en-us/azure/governance/policy/overview (read 10 Nov 2026) - Google Cloud: Test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) - Google Cloud: Policy Simulator for Organization Policy: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) - Google Cloud: Dry run mode for service perimeters: https://cloud.google.com/vpc-service-controls/docs/dry-run-mode (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/testing-a-guardrail-before-you-apply-it/ --- # Roll a guardrail out in stages, keep a dated exception path, and read the denied calls every day Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS, Azure and Google Cloud each document a staged path: test OU first for SCPs and RCPs, DoNotEnforce or Enroll for Azure Policy, dry-run for Google organization policy. This guide sets out the stages, how to handle exceptions and break-glass access, and how to monitor denied calls in CloudTrail and its equivalents. ## Stage the rollout by scope, smallest first AWS states it strongly recommends that you do not attach an SCP to the root of your organization without thorough testing, and suggests an OU that you move accounts into one at a time, or in small numbers. For RCPs the advice is to start with individual test accounts, move up to OUs low in the hierarchy, then work up the structure. The same ordering works elsewhere. Azure assignments and Google policies attach to a management group, folder or project, and a lower scope takes the change first. Because AWS inheritance means a deny near the root applies to everything below it, the root is the last stop and not the first. A pilot subscription or project is the cost of going slowly, and it is small. ## Learn what is used before you restrict it Before attaching an SCP, look at what the accounts do. AWS names two sources: IAM service last accessed data, which shows which services an account used, and CloudTrail logging at the API level. A deny list for three services is safe only if the logs show none of them in use. Last accessed data shows what was used, not what will be needed next quarter, so ask the account owners as well. For an allow list the work is larger, because every service in use must be named. AWS notes that removing FullAWSAccess without a replacement fails all actions from member accounts. Build the allow list from observed use, add a margin for global services, and test it in the OU before it goes near the root. Keep the allow list in version control beside the SCP. ## Plan exceptions and break-glass before the first deny Most library SCPs exempt one named role with a condition. That role is the exception path, so decide who can assume it, how it is logged and how it is reviewed, before you attach the policy. A guardrail without an exception path gets removed in the first incident. Name the role in the policy exactly as it appears in the account, since a typo or an unreplaced placeholder leaves the exception empty. Azure has a built-in version. An exemption with a Waiver category and an expiresOn date lets one resource or hierarchy stay out of the policy for a stated time, and the object stays on record after it lapses. Google organization policy can scope a rule by tag, for example enforcing a constraint only on resources tagged environment=development. Set an expiry on each exception and review them together. ## Use the report-only phase of each cloud On Azure, assign with DoNotEnforce, read the compliance results, then switch to the default mode. When the rollout must go subscope by subscope, the Enroll mode enforces only on resources that are enrolled. On Google Cloud, set the organization policy in dry-run mode so violations are audit logged and nothing is denied, and use the dry run mode of a VPC Service Controls perimeter the same way. A report-only switch for SCPs: none found on 2026-11-10 in the pages read, so the staged OU and the simulator take its place. Control Tower detective controls can report on a related condition in parallel. Control Tower detective controls report a status of clear or in violation during the same period, so you can watch the condition an SCP will block before it is blocked. ## Watch denied calls every day during the rollout On AWS, AWS documents CloudTrail as the place to review Access Denied errors after an RCP is attached. On Azure, a deny returns a 403 and audit effects write to the activity log. On Google Cloud, dry-run violations land in audit logs, and Google documents Cloud Audit Logs for monitoring access patterns across VPC Service Controls perimeters. Name who reads the logs each morning during the rollout. Set a review owner and a rollback rule before the first attachment, for example detach if production errors rise above an agreed count. Keep the previous policy JSON, since disabling the SCP policy type in an AWS root detaches every SCP and the attachments are not recoverable. [Testing a guardrail first](/guides/testing-a-guardrail-before-you-apply-it/) lists the checks that come before this stage. Write the rollback rule down before the attachment. ## A staged plan, cloud by cloud The table puts the documented stages in one place. The names change by cloud, and the order does not: report first, smallest scope second, wider scope third. Use it as the skeleton of the change request. Fill in the owner and the date for each cell before the change is approved, and keep the table with the change record afterward. | Stage | AWS | Azure | Google Cloud | | --- | --- | --- | --- | | 1. Report | Service last accessed data, CloudTrail | DoNotEnforce | Dry-run mode | | 2. Smallest scope | Test OU or test account | Pilot subscription or Enroll | Project or folder | | 3. Exceptions | Privileged role in the policy condition | Dated Waiver exemption | Tag condition on a rule | | 4. Wider scope | Parent OU, then root | Management group | Parent folder, then organization | ## Related pages - [AWS guardrails: Bedrock Guardrails vs account guardrails](/aws-guardrails/) - [AWS Control Tower guardrails: now called controls](/aws-control-tower-guardrails/) - [Multi cloud governance: three hierarchies, three mechanisms](/multi-cloud-governance/) - [All of them](/guides/) ## Sources - AWS Organizations: Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - AWS Organizations: Resource control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - Azure Policy: Assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) - Azure Policy: Exemption structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/exemption-structure (read 10 Nov 2026) - Google Cloud: Test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) - Google Cloud: VPC Service Controls overview: https://cloud.google.com/vpc-service-controls/docs/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guides/rolling-out-guardrails-without-breaking-production/ --- # AWS guardrails means two things: Amazon Bedrock Guardrails for model text, and account-level controls such as SCPs Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS uses the word guardrails for two unrelated features. Amazon Bedrock Guardrails filters prompts and model responses. Account-level guardrails, meaning service control policies, resource control policies, declarative policies and Control Tower controls, cap what accounts can do. This page separates them, then covers the account kind, read 16 Nov 2026. ## Amazon Bedrock Guardrails filters what goes into and out of a model AWS documentation says Amazon Bedrock Guardrails "provides configurable safeguards to help you build safe generative AI applications" and detects and filters undesirable content in user inputs and model responses. It sits at the application layer, next to a model call, and does not control which AWS accounts, Regions or services a person can use. The documentation lists these safeguards: content filters (Hate, Insults, Sexual, Violence, Misconduct and Prompt Attack), denied topics, word filters, sensitive information filters that block or mask personal data, contextual grounding checks, and Automated Reasoning checks. A guardrail has a working draft and a built-in test window. After testing, the owner creates a version and passes the guardrail ID and version on an inference call, or calls the ApplyGuardrail API without invoking a model (AWS documentation, read 16 Nov 2026). Everything below is about the other meaning. A search for model input filtering belongs with the Bedrock documentation, and this site does not cover it further. ## Account-level AWS guardrails are policies that cap what accounts can do AWS Organizations documentation describes a service control policy as defining "a permission guardrail, or sets limits" on what IAM users and roles can do. The library on this site records 40 AWS entries across four mechanisms. Each entry states what it blocks and what it breaks. | Mechanism | What it does | Library entries | | --- | --- | --- | | [Service control policy](/mechanisms/scp/) | Limits the actions available to principals in member accounts | 27 | | [Resource control policy](/mechanisms/rcp/) | Limits access to resources in member accounts, whoever the caller is | 6 | | [Declarative policy](/mechanisms/declarative-policy/) | Sets a service configuration baseline in the service control plane | 5 | | [Permissions boundary](/mechanisms/permissions-boundary/) | Caps what an IAM user or role can be granted | 2 | ## Control Tower controls bundle those mechanisms under one console AWS Control Tower implements preventive controls with service control policies and resource control policies, detective controls with AWS Config rules, and proactive controls with CloudFormation hooks. AWS says it is moving from the word guardrail to the word control and treats the two as synonyms. [AWS Control Tower guardrails](/aws-control-tower-guardrails/) covers the three types and their limits. The two feature families differ in who writes the rule. In Control Tower the rule comes from an AWS-owned catalog. With a plain service control policy the rule is a JSON document you write, such as [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/). ## One library entry connects both meanings: a declarative policy that enforces a Bedrock guardrail AWS documents a declarative policy that references a versioned Bedrock guardrail from the management account and applies it to inference calls in a Region. The library records the conditions: the guardrail must be owned by the management account, must have a non-DRAFT version, and needs a resource-based policy that lets organization members call ApplyGuardrail. See [Enforce an Amazon Bedrock guardrail on model inference](/guardrails/aws-declarative-bedrock-guardrail-enforcement/). A way to test that policy before it ships: none found on 10 Nov 2026 in the AWS pages read. For the general method, see [testing a guardrail before you apply it](/guides/testing-a-guardrail-before-you-apply-it/). ## The quickest test for which meaning you need is the object being restricted Account-level guardrails are the subject of this site. The [library](/library/) lists all 100 entries for AWS, Azure and Google Cloud, and the [method](/method/) explains how platforms that manage them are ranked. - If the thing restricted is a prompt or a model response, you need Bedrock Guardrails. - If the thing restricted is an API action, a Region, an encryption setting or a resource policy in an AWS account, you need an account-level guardrail. - If the thing restricted is a CloudFormation stack before it deploys, you need a Control Tower proactive control or an engine such as [CloudFormation Guard](/engines/cloudformation-guard/). ## Related pages - [AWS Control Tower guardrails: now called controls](/aws-control-tower-guardrails/) - [Multi cloud governance: three hierarchies, three mechanisms](/multi-cloud-governance/) - [What is cloud governance? The security case](/what-is-cloud-governance/) - [All of them](/guides/) ## Sources - AWS: Detect and filter harmful content by using Amazon Bedrock Guardrails: https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails.html (read 16 Nov 2026) - AWS Organizations: Amazon Bedrock policy syntax: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_bedrock_syntax.html (read 10 Nov 2026) - AWS Organizations: service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) - AWS Control Tower: how controls work: https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/aws-guardrails/ --- # AWS Control Tower guardrails are now called controls, and they come in three types: preventive, detective and proactive Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS Control Tower documentation says it is moving from the term guardrail to the term control and treats them as synonyms. Preventive controls use SCPs and RCPs, detective controls use AWS Config rules, and proactive controls use CloudFormation hooks. Each type has a different reach, read 16 Nov 2026. ## A control is one plain-language rule applied to an entire OU AWS documentation defines a control as "a high-level rule that provides ongoing governance for your overall AWS environment" and says it applies to an entire organizational unit. Every account in the OU is subject to it. AWS notes that the old term, guardrail, still appears in the console, blogs and videos, and calls the two terms synonymous. Controls fall into three groups by guidance. Mandatory controls apply to every OU, are applied when the landing zone is set up, and cannot be deactivated. Strongly recommended and elective controls can be changed at any time from the console or the APIs. ## Each control type is a different AWS mechanism with a different reach The reach column is the part buyers miss. A preventive control covers accounts that are not enrolled in Control Tower. A detective control does not. AWS also states that proactive controls cannot evaluate secure strings and secrets, because those values are not resolved before they reach the hook. | Type | Implemented with | When it acts | Status values | Documented limit | | --- | --- | --- | --- | --- | | Preventive | [Service control policies](/mechanisms/scp/) and [resource control policies](/mechanisms/rcp/) | Denies the action | enforced or not enabled | Applies to all member accounts under the OU, enrolled and unenrolled | | Detective | AWS Config rules | After the resource exists | clear, in violation or not enabled | Applies to enrolled accounts only; only in supported Regions | | Proactive | CloudFormation hooks (preCreate and preUpdate) | Before a stack operation creates or updates a resource | Not stated in the page read | May not affect requests made through the console, APIs, SDKs or other IaC tools | ## Three mandatory controls are detective, not preventive AWS states that the mandatory controls have preventive behavior except three that are detective. The three are Detect Public Read Access Setting for Log Archive, Detect Public Write Access Setting for Log Archive, and Detect whether shared accounts under the Security organizational unit have AWS CloudTrail or CloudTrail Lake enabled. The last one is marked for removal in landing zone 4.0. The management account is exempt on purpose. AWS says the root user and administrators in the management account can perform work that controls would otherwise deny, to keep that account from becoming unusable. Actions there are still logged in the log archive account. ## Configurable controls take parameters, including exempted principals and actions AWS shows that a control can be enabled on an OU through CloudFormation with the resource type AWS::ControlTower::EnabledControl, a control identifier and a target OU identifier. Its example passes three parameters: AllowedRegions, ExemptedPrincipalArns (a role such as arn:aws:iam::*:role/ReadOnly) and ExemptedActions (for example logs:StartQuery). These parameters are the native exception route for catalog controls. They are set per OU, so an exemption covers every account in that OU. The page read gives no expiry field for them: none found on 10 Nov 2026. [Guardrail exceptions and waivers](/guardrail-exceptions-and-waivers/) compares this with the Azure and Google Cloud routes. ## Where Control Tower controls stop and plain SCPs begin Control Tower controls come from an AWS-owned catalog. A rule outside the catalog, such as [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/), is written as a service control policy and attached to the OU yourself. The mechanism pages explain evaluation and limits, and [how SCP evaluation works](/guides/how-scp-evaluation-works/) covers the allow-list and deny-list models. To compare Control Tower with other products on safe rollout, see the [AWS Control Tower platform page](/platforms/aws-control-tower/). For the general split between the three types, see [preventive vs detective controls](/guides/preventive-vs-detective-controls/) and the [proactive control](/glossary/proactive-control/) glossary entry. ## Related pages - [Multi cloud governance: three hierarchies, three mechanisms](/multi-cloud-governance/) - [What is cloud governance? The security case](/what-is-cloud-governance/) - [Cloud governance framework: six layers for security](/cloud-governance-framework/) - [All of them](/guides/) ## Sources - AWS Control Tower: how controls work: https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html (read 10 Nov 2026) - AWS Control Tower: about controls: https://docs.aws.amazon.com/controltower/latest/userguide/controls.html (read 16 Nov 2026) - AWS Control Tower: preventive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html (read 10 Nov 2026) - AWS Control Tower: detective controls: https://docs.aws.amazon.com/controltower/latest/controlreference/detective-controls.html (read 10 Nov 2026) - AWS Control Tower: proactive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) - AWS Control Tower: mandatory controls: https://docs.aws.amazon.com/controltower/latest/controlreference/mandatory-controls.html (read 10 Nov 2026) - AWS Control Tower: enable controls with CloudFormation: https://docs.aws.amazon.com/controltower/latest/userguide/enable-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/aws-control-tower-guardrails/ --- # Multi cloud governance for security means running three hierarchies and three policy mechanisms side by side Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS, Azure and Google Cloud each organize accounts in a hierarchy and attach guardrails to it: AWS Organizations with SCPs and RCPs, Azure management groups with Azure Policy, Google Cloud folders with organization policy. The vocabulary, the dry-run options and the exception models differ in each, as read on 16 Nov 2026. ## Each cloud attaches guardrails to a different container AWS documentation says Organizations lets you group accounts and apply policies for governance, and it calls accounts natural boundaries for permission, security, costs and workloads. Microsoft says management groups give a governance scope above subscriptions, and that a policy applied to a management group reaches all nested groups, subscriptions and resources. Google says its hierarchy of organizations, folders and projects carries inherited access control and organization policies. | Question | AWS | Azure | Google Cloud | | --- | --- | --- | --- | | Container that carries the guardrail | Root, OU or account | Management group or subscription | Organization, folder or project | | Main preventive mechanisms | [SCP](/mechanisms/scp/), [RCP](/mechanisms/rcp/), [declarative policy](/mechanisms/declarative-policy/) | [Azure Policy](/mechanisms/azure-policy/), [deny assignments](/mechanisms/rbac-deny-assignment/) | [Organization policy](/mechanisms/org-policy/), [IAM deny](/mechanisms/iam-deny/), [VPC Service Controls](/mechanisms/vpc-service-controls/) | | Library entries on this site | 40 | 30 | 30 | | Documented way to try before enforcing | Test in a separate OU; IAM policy simulator for SCPs | enforcementMode DoNotEnforce | Organization policy in dry-run mode | | Depth limit documented | None found in the pages read | Six levels, excluding root and subscription | None found in the pages read | ## The same intent needs a different object in each cloud Restricting locations shows the gap. AWS uses a service control policy with an aws:RequestedRegion condition, as in [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/). Azure uses a policy definition, as in [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/). Google uses a constraint, as in [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/). The failure modes differ too. The AWS example breaks global services missing from its exemption list. The Azure definition excludes resource groups and resources in the global region. The Google constraint treats a multi-region deny as separate from its sub-locations, so a deny of the us multi-region still allows us-east1. ## Inheritance works downward in all three, with different merge rules AWS documentation shows that when SCPs are stacked, the more restrictive root policy takes precedence, so the effective permission is the intersection of every level. Microsoft recommends limiting the number of Azure Policy assignments at the root management group, because inherited policies are harder to debug in lower groups. Google lets a child policy merge with its parent or replace it, depending on the inheritFromParent setting. The practical result is that a guardrail written high in one cloud can be loosened lower in another. In Google Cloud and Azure a lower scope can override or exempt. In AWS an SCP can only narrow what is allowed, so loosening needs an exception condition in the policy itself or a change higher up. ## Dry-run support is the least even feature across the three Microsoft documents DoNotEnforce as a way to evaluate an assignment before enforcement. Google documents dry-run mode, where violations are audit logged but not blocked, and limits it to certain constraint types. For SCPs, AWS documents the IAM policy simulator and recommends a small OU first. A dry-run mode for SCPs: none found in the AWS pages read on 10 Nov 2026. This is the gap the [Simulation Coverage Matrix](/simulation/) records for the platforms that sit on top of the three clouds. The [multi-cloud situation ranking](/situations/multi-cloud/) weights it, and the [method](/method/) lists the weights. ## A cross-cloud rulebook is a table of equivalents, not one policy No page read on 16 Nov 2026 describes a single policy file that all three clouds enforce natively. Teams that want one rulebook keep a list of intents (no public storage, approved Regions, no long-lived keys) and map each to the native object per cloud, or add an external engine; see [policy as code tools](/policy-as-code-tools/) for the engines. ## Related pages - [What is cloud governance? The security case](/what-is-cloud-governance/) - [Cloud governance framework: six layers for security](/cloud-governance-framework/) - [Guardrail exceptions and waivers in AWS, Azure and Google](/guardrail-exceptions-and-waivers/) - [All of them](/guides/) ## Sources - AWS Organizations: what is AWS Organizations: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_introduction.html (read 16 Nov 2026) - Microsoft: management groups overview: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 16 Nov 2026) - Google Cloud: resource hierarchy: https://cloud.google.com/resource-manager/docs/cloud-platform-resource-hierarchy (read 16 Nov 2026) - Microsoft: management groups in Azure landing zones: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/resource-org-management-groups (read 16 Nov 2026) - Microsoft: Azure Policy assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 16 Nov 2026) - Google Cloud: test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 16 Nov 2026) - AWS Organizations: SCP strategies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_strategies.html (read 16 Nov 2026) - Google Cloud: hierarchy evaluation: https://cloud.google.com/resource-manager/docs/organization-policy/understanding-hierarchy (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/multi-cloud-governance/ --- # Cloud governance, in its security sense, is a set of rules the cloud itself enforces on every account Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cloud governance for security is the practice of setting rules once, high in the account hierarchy, so AWS, Azure or Google Cloud applies them to every account beneath. This site covers the preventive layer: 100 documented guardrails, read 16 Nov 2026. Cost governance and posture reporting are separate subjects. ## Cloud security governance sets limits at the top and lets the cloud apply them The definition used on this site is narrow. Cloud governance is the set of policies attached at the top of an account hierarchy, such as an AWS organization root, an Azure management group or a Google Cloud organization, that limit what the accounts below can do. The cloud evaluates them on every request, so no team has to remember them. Microsoft frames the same idea as a replacement for review meetings. Its Cloud Adoption Framework says Azure Policy "replaces such reviews with automated guardrails and adherence audits" (Microsoft documentation, read 16 Nov 2026). AWS says its Organizations service lets you apply policies for governance across accounts, and calls accounts natural boundaries for permission and security. ## This site covers the guardrail layer, which is one of three jobs governance is asked to do People who search for cloud governance usually mean one of three jobs: spending limits, a report of what is misconfigured, or rules that stop a bad change. This site covers the third. Cost governance is out of scope, and a report of misconfigurations is a different tool category than the policies that prevent them. The library holds 100 entries: 40 for AWS, 30 for Azure and 30 for Google Cloud. The largest risk classes are [AI services](/risk/ai-services/) with 19 entries, [identity and privilege](/risk/identity-and-privilege/) with 14 and [encryption and keys](/risk/encryption-and-keys/) with 13. Each entry names what it blocks and what it breaks, with the source of both. | Cloud | Where the rule is attached | Rule objects in the library | | --- | --- | --- | | AWS | Organization root, OU or account | [SCP](/mechanisms/scp/), [RCP](/mechanisms/rcp/), [declarative policy](/mechanisms/declarative-policy/), [permissions boundary](/mechanisms/permissions-boundary/) | | Azure | Management group, subscription or resource group | [Azure Policy](/mechanisms/azure-policy/), [deny assignment](/mechanisms/rbac-deny-assignment/), [resource lock](/mechanisms/resource-lock/) | | Google Cloud | Organization, folder or project | [Organization policy](/mechanisms/org-policy/), [IAM deny](/mechanisms/iam-deny/), [principal access boundary](/mechanisms/principal-access-boundary/), [VPC Service Controls](/mechanisms/vpc-service-controls/) | ## Governance is judged by what it stops, what it breaks and how fast it can be changed A program is only as good as its three properties. It must stop the thing it names. It must break little else, and it must say what it breaks. And a rule must be changeable without a week of testing, which is why the ranking on this site weights safe enforcement and one change per class of risk at 30 each. The [method](/method/) publishes all seven weights. A single account is not governed by this definition, because there is nothing above it to attach a rule to. The case starts at two accounts: one holding the rule, one receiving it. ## A governance rule is a denial with a cost, so the cost is part of the definition A guardrail answers a request with a denial. That is its value and its risk. The [glossary](/glossary/guardrail/) defines a guardrail as a policy that sets limits without granting any permission. A rule that blocks the wrong request stops legitimate work, which is why every library entry lists the breakage as well as the benefit. Take [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/). It has no exception condition, so it applies to every user, role and root user in the affected member accounts. AWS documents that service control policies do not affect the management account. Governance in this sense is writing such a rule, knowing where it stops, and testing it before it ships. ## The next steps are a framework, a rollout plan and a choice of tools For the layers a program needs, read [a cloud governance framework for security](/cloud-governance-framework/). For the spread of rules across three clouds, read [multi cloud governance](/multi-cloud-governance/). For how platforms are ranked on getting a rule into production safely, read the [method](/method/). ## Related pages - [Cloud governance framework: six layers for security](/cloud-governance-framework/) - [Guardrail exceptions and waivers in AWS, Azure and Google](/guardrail-exceptions-and-waivers/) - [Landing zone guardrails for AWS: the Control Tower baseline](/landing-zone-guardrails-aws/) - [All of them](/guides/) ## Sources - Microsoft: Azure governance design area: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/governance (read 16 Nov 2026) - AWS Organizations: what is AWS Organizations: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_introduction.html (read 16 Nov 2026) - AWS Organizations: service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) - Google Cloud: resource hierarchy: https://cloud.google.com/resource-manager/docs/cloud-platform-resource-hierarchy (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/what-is-cloud-governance/ --- # A cloud governance framework for security has six layers, from account structure to evidence Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A security-focused cloud governance framework has six layers: account structure, baseline preventive guardrails, detective controls, exceptions, a test step, and evidence. Each cloud supplies native objects for every layer. In the library, 26 of 100 entries show no native test step found on 10 Nov 2026, mostly in Google Cloud. ## Each layer answers one question and maps to a native object The table is a summary of documentation read on 16 Nov 2026. Layers 3 and 6 are the thinnest in the library, because this site covers preventive guardrails; detective mechanisms are named only where a cloud puts them next to a preventive one. | Layer | Question it answers | AWS | Azure | Google Cloud | | --- | --- | --- | --- | --- | | 1. Structure | Where does a rule attach? | Organization, OUs, accounts | Management groups, subscriptions | Organization, folders, projects | | 2. Baseline preventive | What is never allowed? | [SCP](/mechanisms/scp/), [RCP](/mechanisms/rcp/) | [Azure Policy](/mechanisms/azure-policy/) with deny | [Organization policy](/mechanisms/org-policy/), [IAM deny](/mechanisms/iam-deny/) | | 3. Detective | What drifted anyway? | AWS Config rules (Control Tower detective controls) | Azure Policy with audit effect | Dry-run violations in audit logs; other detective tooling not read for this page | | 4. Exceptions | Who may break a rule, and until when? | Principal tag or ARN conditions; Control Tower exempted parameters | Policy exemptions with expiresOn | Folder or project override, tag conditions | | 5. Test | What does the rule break? | Test OU, policy simulator for SCPs | enforcementMode DoNotEnforce | Dry-run mode | | 6. Evidence | Can you show the rule held? | Denies logged in CloudTrail | Compliance reporting and auditing | Audit logs of dry-run and enforced violations | ## Structure comes first because it decides how many rules you need Microsoft recommends keeping the management group hierarchy reasonably flat, ideally no more than three to four levels, and states a limit of six levels of depth. It also recommends limiting the number of Azure Policy assignments at the root management group scope. AWS documents a 10,240 character ceiling per service control policy and a cap on how many SCPs attach to a root, OU or account. Those numbers shape every later layer. See [landing zone guardrails for AWS](/landing-zone-guardrails-aws/) and [landing zone guardrails for Azure](/landing-zone-guardrails-azure/) for the structures each provider recommends. ## The test layer is where the three clouds differ most The library records, for each of 100 entries, the native way to test before enforcing. For 26 of them the answer is "none found on 10 Nov 2026". Twenty-three are Google Cloud entries, and all 23 are legacy managed constraints, which Google documents as outside Policy Simulator and mostly outside dry-run mode. Two are Azure entries and one is AWS. That makes the test layer the main reason an outside tool earns a place in the framework. The [Simulation Coverage Matrix](/simulation/) shows which platforms add a test step the clouds lack, and [testing a guardrail before you apply it](/guides/testing-a-guardrail-before-you-apply-it/) walks through the native options. ## A short order of work follows from the layers The order is this page's own recommendation. It follows the documented limits above and is not taken from a single provider's guidance. - Fix the structure first: one baseline attachment point per cloud, and a sandbox group with a looser set. - Add the small number of rules that must never be broken, and read the breakage recorded for each entry before attaching it. - Put every new rule through the cloud's test step, or a platform that adds one, before it is enforced. - Give every exception an owner and an end date in whatever field or record the cloud allows. - Keep the log that shows denials, so that the rule can be shown to have held. ## Exceptions and evidence are the layers audits ask about A guardrail with a hard-coded exception is a guardrail with a permanent hole. Of the three clouds, only Azure documents an expiry field on its exception object in the pages read; see [guardrail exceptions and waivers](/guardrail-exceptions-and-waivers/). For rollout order across layers, see [rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/). For evidence, AWS says a prevented action "is denied and logged in CloudTrail" and the resource is logged in AWS Config (AWS Control Tower documentation, read 10 Nov 2026). Microsoft lists compliance reporting and auditing as a deployment consideration. Google documents that dry-run violations are audit logged without blocking. ## Related pages - [Guardrail exceptions and waivers in AWS, Azure and Google](/guardrail-exceptions-and-waivers/) - [Landing zone guardrails for AWS: the Control Tower baseline](/landing-zone-guardrails-aws/) - [Azure landing zone guardrails: policy by management group](/landing-zone-guardrails-azure/) - [All of them](/guides/) ## Sources - Microsoft: Azure governance design area: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/governance (read 16 Nov 2026) - Microsoft: management groups in Azure landing zones: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/resource-org-management-groups (read 16 Nov 2026) - Microsoft: Azure Policy assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 16 Nov 2026) - Microsoft: Azure Policy exemption structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/exemption-structure (read 16 Nov 2026) - Google Cloud: test organization policies in dry-run mode: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 16 Nov 2026) - AWS Control Tower: how controls work: https://docs.aws.amazon.com/controltower/latest/userguide/how-controls-work.html (read 10 Nov 2026) - AWS Organizations: quotas and limits: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - Microsoft: Azure Policy audit effect: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-audit (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/cloud-governance-framework/ --- # Only Azure documents an exception object with an expiry date; AWS and Google Cloud exceptions need an expiry built around them Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Azure Policy exemptions carry a category, an optional expiresOn date and free-form metadata. AWS exceptions are conditions inside the policy, such as a principal tag in an SCP. Google Cloud exceptions are overrides at a folder or project. An expiry field for the last two: none found on 16 Nov 2026. ## AWS exceptions are conditions inside the policy, so ending one means editing the policy The AWS sample service control policy [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) denies IAM user creation, access keys and login profiles unless the caller is a named role or carries a principal tag. The exception is the Condition block below, copied from the aws-samples repository (MIT-0 license), with indentation removed. - `"Condition": {` - `"ArnNotLike": { "aws:PrincipalArn": "arn:aws:iam::*:role/[PRIVILEGED_ROLE]" },` - `"StringNotLike": { "aws:PrincipalTag/IAMUserManagementException": "true" }` - `}` ## AWS Control Tower adds exemption parameters per OU, with no expiry field found AWS Control Tower documentation shows configurable controls that accept ExemptedPrincipalArns and ExemptedActions parameters, set when the control is enabled on an OU. The pages read list no expiry or approval field for either parameter: none found on 10 Nov 2026. An exception through the tag route lasts as long as the tag stays on the principal. The [Control Tower controls page](/aws-control-tower-guardrails/) covers the catalog. ## Azure exemptions are separate objects with a category, an expiry and a record Microsoft documents a policy exemption as a child object on the resource hierarchy or resource it covers. Its fields include a policy assignment ID, optional policy definition reference IDs (to exempt only some definitions in an initiative), an exemption category, an optional expiresOn date and a free-form metadata block. Microsoft gives requestedBy, approvedBy, approvedOn and ticketRef as example metadata keys. The two categories are Mitigated (the policy intent is met another way) and Waiver (non-compliance is temporarily accepted). When expiresOn passes, the exemption object is kept for record-keeping but is no longer honored. Microsoft says exempt resources count toward overall compliance but are not evaluated, or hold a temporary waiver. See [Azure Policy as a guardrail](/guides/azure-policy-as-a-guardrail/) and the [policy exemption](/glossary/policy-exemption/) entry. ## Google Cloud exceptions are overrides at a lower node or tag conditions, with a delay Google documents that a project can override an inherited boolean organization policy by setting a policy that disables enforcement for that project. The YAML Google gives for that override is shown below, with indentation removed. - `name: projects/PROJECT_ID/policies/CONSTRAINT_NAME` - `spec:` - `rules:` - `- enforce: false` | Cloud | Exception object | Scope | Expiry field | Record of the reason | | --- | --- | --- | --- | --- | | AWS SCP | Condition on principal ARN or tag | The attached OU, account or root | None found in the SCP samples read | Kept outside the policy | | AWS Control Tower | ExemptedPrincipalArns and ExemptedActions | The OU | None found on 10 Nov 2026 | None found on 10 Nov 2026 | | Azure Policy | policyExemptions child object | Resource hierarchy or one resource | expiresOn, optional | metadata block and exemptionCategory | | Google Cloud | Override at folder or project; tag-conditional enforcement | The folder or project | None found on 16 Nov 2026 | None found on 16 Nov 2026 | ## Expiry for AWS and Google Cloud has to be built, and timing matters Google states that changes to organization policies can take up to 15 minutes to be fully enforced, so an exception that is lifted is not lifted at once. Google also documents tag-based conditions, which allow or deny a policy depending on whether a resource has a specific tag. Tags on organization, folder or project resources work with any constraint type, and tags on service resources such as Compute Engine instances work with managed or custom constraints. A review job that lists tagged principals or project-level overrides and removes stale ones is the usual stand-in for an expiry field. For how to stage exceptions during a rollout, read [rolling out guardrails without breaking production](/guides/rolling-out-guardrails-without-breaking-production/), and see the [library](/library/) for the breakage each guardrail records. ## Related pages - [Landing zone guardrails for AWS: the Control Tower baseline](/landing-zone-guardrails-aws/) - [Azure landing zone guardrails: policy by management group](/landing-zone-guardrails-azure/) - [What is policy as code? Definition and where it runs](/what-is-policy-as-code/) - [All of them](/guides/) ## Sources - AWS samples: Deny-critical-IAM-user-actions.json (MIT-0): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-critical-IAM-user-actions.json (read 10 Nov 2026) - AWS Control Tower: enable controls with CloudFormation: https://docs.aws.amazon.com/controltower/latest/userguide/enable-controls.html (read 10 Nov 2026) - Microsoft: Azure Policy exemption structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/exemption-structure (read 16 Nov 2026) - Google Cloud: using constraints: https://cloud.google.com/resource-manager/docs/organization-policy/using-constraints (read 16 Nov 2026) - Google Cloud: scope organization policies with tags: https://cloud.google.com/resource-manager/docs/organization-policy/tags-organization-policy (read 16 Nov 2026) - Google Cloud: hierarchy evaluation: https://cloud.google.com/resource-manager/docs/organization-policy/understanding-hierarchy (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrail-exceptions-and-waivers/ --- # An AWS landing zone starts with mandatory controls on every OU, then adds guardrails OU by OU Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. An AWS Control Tower landing zone applies mandatory controls to every organizational unit when it is set up, and AWS says they cannot be deactivated. Landing zone 4.0 stopped deploying four CloudTrail controls. This page lists the baseline and the library guardrails that match it, as read on 16 Nov 2026. ## AWS defines a landing zone as a recommended starting point for accounts, network and security AWS documentation defines a landing zone as "a cloud environment that offers a recommended starting point, including default accounts, account structure, network and security layouts". It says Control Tower acts as an orchestration layer over other AWS services to implement the AWS multi-account recommendations, and that one account is not enough for a well-architected environment. AWS lists the reasons for multiple accounts: security controls, isolation, many teams, data isolation, business process, billing and quota allocation. Each is a reason to attach a different guardrail to a different OU. ## Mandatory controls are the fixed baseline, and they use the same exception pattern you would write AWS says mandatory controls are owned by Control Tower, apply to every OU in the landing zone (with some exceptions for the Security OU), are applied by default at setup and cannot be deactivated, because they protect Control Tower resources. The reference page prints their policy text. One of them, Disallow changes to Amazon SNS subscriptions and topics managed by AWS Control Tower, uses Sid CTSNSPV1. It denies every action except a list (confirm, get, list, publish, subscribe, unsubscribe and tagging actions) on topics named aws-controltower-CentralizedLoggingNotifications*, and exempts one principal with an ArnNotLike condition on the role AWSControlTowerExecution. That is the same NotAction plus exempt-role shape that the sample SCPs in the [library](/library/) use. ## Landing zone 4.0 stopped deploying four CloudTrail controls and removed others AWS states that from landing zone 4.0 it no longer deploys four controls that protect account trails created in landing zone 2.9 and below: Disallow Configuration Changes to CloudTrail, Integrate CloudTrail Events with Amazon CloudWatch Logs, Enable CloudTrail in All Available Regions, and Enable Integrity Validation for CloudTrail Log File. They stay in force on landing zone versions below 4.0 and are deleted on upgrade. Two AWS Config aggregator controls are also removed for customers who upgrade with Config integration enabled. If your baseline was written from an older landing zone, the controls on your OUs and the controls in the current catalog may differ. ## Seven library entries cover the same ground as the baseline Each linked page records what the entry blocks, what it breaks and how AWS says to test it. The [SCP mechanism page](/mechanisms/scp/) lists the evaluation limits. | Baseline goal | Library entry | Mechanism | | --- | --- | --- | | Keep accounts in the organization | [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) | SCP | | Keep audit trails intact | [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) | SCP | | Keep configuration recording on | [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/) | SCP | | Keep detection on | [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) | SCP | | Keep network logs | [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) | SCP | | Limit Regions | [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/) | SCP | | Encrypt in transit | [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/) | RCP | ## AWS recommends four OUs, and guardrails attach to each one separately AWS also recommends at least two accounts, Production and Staging, with distinct controls and policies, governed as separate OUs. Because a control applies to a whole OU, the OU layout decides how finely a guardrail can be aimed. Control Tower supports nested OUs, so a baseline can sit on a parent and exceptions on a child. | OU | Who creates it | Purpose in AWS's description | | --- | --- | --- | | Security | Control Tower, as the first required OU | Shared security accounts; mandatory controls have some exceptions here | | Sandbox | Control Tower, if you select it | Software development; for example a fixed spending limit or no link to the production network | | Infrastructure | You | Shared services and networking accounts | | Workloads | You | Accounts that run workloads | ## Preventive controls reach unenrolled accounts, and detective controls do not AWS states that preventive controls enabled on an OU registered with Control Tower apply to all member accounts under the OU, enrolled and unenrolled, while detective controls apply to enrolled accounts only. Preventive controls enabled on an OU higher in the tree also apply to unregistered OUs below it. Place the baseline high and the exceptions low. See [AWS Control Tower guardrails](/aws-control-tower-guardrails/) for the three control types. ## Related pages - [Azure landing zone guardrails: policy by management group](/landing-zone-guardrails-azure/) - [What is policy as code? Definition and where it runs](/what-is-policy-as-code/) - [Policy as code examples: Rego, Sentinel, Kyverno, Guard](/policy-as-code-examples/) - [All of them](/guides/) ## Sources - AWS Control Tower: multi-account strategy for your landing zone: https://docs.aws.amazon.com/controltower/latest/userguide/aws-multi-account-landing-zone.html (read 10 Nov 2026) - AWS Control Tower: mandatory controls: https://docs.aws.amazon.com/controltower/latest/controlreference/mandatory-controls.html (read 10 Nov 2026) - AWS Control Tower: preventive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html (read 10 Nov 2026) - AWS Control Tower: detective controls: https://docs.aws.amazon.com/controltower/latest/controlreference/detective-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/landing-zone-guardrails-aws/ --- # An Azure landing zone assigns policy to management groups, with sandboxes held to a less restrictive set Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Microsoft Cloud Adoption Framework guidance for Azure landing zones places Azure Policy assignments on management groups: platform, landing zones, sandboxes and decommissioned. It advises few assignments at the root, a flat hierarchy of three to four levels, and no workload RBAC at management group scope, as read on 16 Nov 2026. ## Microsoft assigns policy at management groups, grouped by workload type Microsoft says to use management groups to aggregate policy and initiative assignments, and to organize subscriptions by workload type such as online, corp, local or sandbox. The reference architecture below is Microsoft's description, read 16 Nov 2026. | Management group | Purpose in Microsoft's description | Policy posture stated | | --- | --- | --- | | Platform (Security, Management, Connectivity, Identity) | Shared platform subscriptions | Specific policies to harden and manage the identity subscription | | Landing zones | Parent of all landing zone groups | Workload-agnostic policies so workloads are secure and compliant | | Online and Corp | Internet-facing and corporate-network workloads | Inherit the landing zone assignments | | Local | Workloads on Azure Local clusters | Different Azure Policy requirements from the other groups | | Sandboxes | Testing and exploration, isolated from corp and online | Less restrictive set of policies | | Decommissioned | Canceled landing zones | Azure deletes them after 30 to 60 days | ## The advice is to assign policy at the group and keep the tree shallow Microsoft also notes that by default any principal in a Microsoft Entra tenant can create management groups, and advises enabling Azure RBAC authorization on the hierarchy to restrict it. - Keep the hierarchy reasonably flat, ideally no more than three to four levels; a tree supports up to six levels of depth, excluding root and subscription. - Limit the number of Azure Policy assignments at the root management group scope, to cut debugging of inherited policies lower down. - Do not assign workload teams permissions through RBAC at management group scope; use management groups for policy assignment instead. - Set a dedicated default management group for new subscriptions, because new subscriptions otherwise land under the tenant root group. - Do not create management groups for production, test and development; separate those into subscriptions in the same group. ## Subscriptions are placed by workload type, and sandboxes are the exception to strict policy Microsoft says to create a sandbox management group so teams can experiment with resources before moving them to production, isolated from development, test and production. It says sandboxes have a less restrictive policy set so that testing and exploration are possible. That is a deliberate gap in the baseline, and it makes the default management group for new subscriptions a security decision: whichever group is the default receives every new subscription until someone moves it. For location-based regulatory needs such as data residency, Microsoft advises building a management group structure based on location instead of one group per region, and it advises against creating groups only to model Azure regions. ## A default policy set exists, and Microsoft says to use it as a starting point Microsoft's governance design area says to determine which policies you need from your business controls or compliance regulations, and to use the policies in the Azure landing zone reference architecture as a starting point. It names the Azure landing zone default policy assignments Bicep module as a way to get a head start, and says change advisory review boards can hinder agility where Azure Policy automated guardrails can replace them. What the module assigns, and how many policies: none found in the pages read on 16 Nov 2026. Read the module itself before relying on it. ## Library entries show what a management group assignment does and where it leaks The library entry [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/) records the main trap: scopes listed in notScopes on an assignment are skipped entirely, and are not evaluated or counted in compliance. The same assignment can be tested first with enforcementMode set to DoNotEnforce, which Microsoft documents for evaluating an assignment before enforcement. Other entries suited to an inherited baseline include [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/) and [Resource locks](/guardrails/azure-lock-cannotdelete-readonly/). The [management group mechanism page](/mechanisms/management-groups/) and the [Azure Policy mechanism page](/mechanisms/azure-policy/) explain the objects. ## Related pages - [What is policy as code? Definition and where it runs](/what-is-policy-as-code/) - [Policy as code examples: Rego, Sentinel, Kyverno, Guard](/policy-as-code-examples/) - [Compliance as code: rule bundles for AWS, Azure and Linux](/compliance-as-code/) - [All of them](/guides/) ## Sources - Microsoft: management groups in Azure landing zones: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/resource-org-management-groups (read 16 Nov 2026) - Microsoft: Azure governance design area: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/ready/landing-zone/design-area/governance (read 16 Nov 2026) - Microsoft: Azure Policy assignment structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/landing-zone-guardrails-azure/ --- # Policy as code means writing rules as versioned, testable text that an engine evaluates, in a pipeline or at the cloud API Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Policy as code is the practice of writing rules in a high-level language, keeping them as text in version control, testing them, and having an engine evaluate them automatically. HashiCorp, Open Policy Agent and AWS each publish an engine for it. The rules can run before deployment, at admission or inside the cloud API, as read on 16 Nov 2026. ## HashiCorp defines policy as code as writing code to manage and automate policies HashiCorp's Sentinel documentation says policy as code "is the idea of writing code in a high-level language to manage and automate policies". It contrasts this with policies set by clicking in a console, which it calls hard to repeat and version, with no system for testing a policy other than trying an action that would violate it. The same page lists five benefits: sandboxing (guardrails for automated systems), codification (logic in code rather than oral tradition), version control, testing, and automation. Open Policy Agent describes itself as a policy engine that "unifies policy enforcement across the stack". AWS describes CloudFormation Guard as a general-purpose policy-as-code evaluation tool. ## A rule can be checked at four points, and each point catches different changes The last column is the verdict of this page. No single point covers everything, so a mature setup uses more than one. The [policy as code tools](/policy-as-code-tools/) page lists the engines by the same fields. | Point | What is checked | Engines in this set | What it misses | | --- | --- | --- | --- | | Before deploy, in CI | A Terraform plan in JSON, a CloudFormation template, a Kubernetes manifest | [Open Policy Agent](/engines/open-policy-agent/), [Sentinel](/engines/hashicorp-sentinel/), [CloudFormation Guard](/engines/cloudformation-guard/) | Anything changed in the console or by API | | At cluster admission | A Kubernetes resource as it is created or updated | [Kyverno](/engines/kyverno/), OPA through Gatekeeper | Anything outside the cluster | | At the cloud API | Every request in the account | [SCP](/mechanisms/scp/), [Azure Policy](/mechanisms/azure-policy/), [organization policy](/mechanisms/org-policy/) | Nothing the policy language cannot express | | After creation, from events or schedules | Live resources | [Cloud Custodian](/engines/cloud-custodian/), AWS Config rules | The window before correction | ## The cloud's own policies are policy as code too, and they are the only ones evaluated at the API A service control policy is a JSON document that can live in a repository and go through review like any other code. So is an Azure Policy definition. Every one of the [100 library entries](/library/) names its source file and license, many from open-source repositories. What differs is the evaluator: the cloud itself, rather than a separate engine reading a copy of the intent. The AWS sample SCP repository, for instance, is MIT-0 licensed, which lets a team copy a policy into its own repository and change it. That is why this site treats the two as layers. The cloud-native object is the one that is hard to bypass. An engine upstream is the one that tells a developer early and with a readable message. [Testing a guardrail before you apply it](/guides/testing-a-guardrail-before-you-apply-it/) covers the cloud-native test options. ## Plan-time checks have a documented blind spot Open Policy Agent states that policies on a Terraform plan cannot see unknown values (computed at apply time), dynamic blocks, or the results of functions not yet evaluated. A plan-time rule that depends on such an attribute passes without checking it. HashiCorp documents a related limit: HCP Terraform policies are checked during Terraform runs, against the plan, and HashiCorp says custom imports are not supported. A change made outside Terraform is never presented to them. HashiCorp also says a policy set holds policies from one framework only. Real examples of rules in five engines are on the [policy as code examples](/policy-as-code-examples/) page, and [compliance as code](/compliance-as-code/) covers rule bundles tied to standards. ## Related pages - [Policy as code examples: Rego, Sentinel, Kyverno, Guard](/policy-as-code-examples/) - [Compliance as code: rule bundles for AWS, Azure and Linux](/compliance-as-code/) - [Policy as code tools: five engines by the same eight fields](/policy-as-code-tools/) - [All of them](/guides/) ## Sources - HashiCorp: policy as code (Sentinel): https://developer.hashicorp.com/sentinel/docs/concepts/policy-as-code (read 16 Nov 2026) - Open Policy Agent: introduction: https://www.openpolicyagent.org/docs (read 16 Nov 2026) - Open Policy Agent: Terraform: https://www.openpolicyagent.org/docs/terraform (read 16 Nov 2026) - AWS: what is CloudFormation Guard: https://docs.aws.amazon.com/cfn-guard/latest/ug/what-is-guard.html (read 16 Nov 2026) - HashiCorp: policy enforcement in HCP Terraform: https://developer.hashicorp.com/terraform/cloud-docs/policy-enforcement (read 16 Nov 2026) - GitHub: kyverno/kyverno: https://github.com/kyverno/kyverno (read 16 Nov 2026) - GitHub: cloud-custodian/cloud-custodian: https://github.com/cloud-custodian/cloud-custodian (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/what-is-policy-as-code/ --- # Policy as code examples in five engines show one rule shape: select a resource, test one attribute, return a verdict Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Five short policy as code examples, each copied from its project's own documentation: a Rego rule for Open Policy Agent, a Sentinel allow list, a Cloud Custodian YAML policy, a Kyverno ClusterPolicy and a CloudFormation Guard rule. Each is attributed with the project license, read 16 Nov 2026, with indentation removed. ## Rego flags a security group whose description mentions HTTP From the Open Policy Agent Terraform guide. The project is Apache License 2.0. The helper that collects resources from the plan is on the same page and omitted here. Lines, in order: - `package terraform.module` - `deny contains msg if {` - `some r` - `desc := resources[r].values.description` - `contains(desc, "HTTP")` - `msg := sprintf("No security groups should be using HTTP. Resource in violation: %v", [r.address])` - `}` ## Sentinel checks an EC2 instance type against an allow list From HashiCorp's page on defining Sentinel policies in HCP Terraform. The license for Sentinel: none found on 16 Nov 2026. The function validate_ec2_instance_types is defined elsewhere on that page. - `allowed_types = [` - `"t2.small",` - `"t2.medium",` - `"t2.large",` - `]` - `main = rule {` - `validate_ec2_instance_types(allowed_types)` - `}` ## Cloud Custodian removes cross-account access from S3 buckets From the Cloud Custodian README. The project is Apache License 2.0. The description lines are omitted. - `policies:` - `- name: s3-cross-account` - `resource: aws.s3` - `region: us-east-1` - `filters:` - `- type: cross-account` - `actions:` - `- type: remove-statements` - `statement_ids: matched` ## Kyverno blocks a Namespace that lacks a purpose label From the Kyverno documentation on validate rules. The project is Apache License 2.0. Comments and the message line are omitted. - `apiVersion: kyverno.io/v1` - `kind: ClusterPolicy` - `metadata:` - `name: require-ns-purpose-label` - `spec:` - `rules:` - `- name: require-ns-purpose-label` - `match:` - `any:` - `- resources:` - `kinds:` - `- Namespace` - `validate:` - `failureAction: Enforce` - `pattern:` - `metadata:` - `labels:` - `purpose: production` ## CloudFormation Guard requires an S3 bucket prefix, KMS encryption and a skip when none exist From the CloudFormation Guard README. The project is Apache License 2.0. - `let s3_buckets = Resources.*[ Type == /S3::Bucket/ ]` - `rule s3_bucket_name_encryption_check when %s3_buckets !empty {` - `%s3_buckets {` - `Properties {` - `BucketName == /^MyCompanyPrefix/` - `BucketEncryption.ServerSideEncryptionConfiguration[*] {` - `ServerSideEncryptionByDefault.SSEAlgorithm IN ["aws:KMS"]` - `}` - `}` - `}` - `}` ## Reuse terms are the project license, and documentation terms were not read separately Each snippet above sits in the project's own documentation or README, and each repository carries the license stated. A separate license for the documentation site itself: none found on 16 Nov 2026 in the pages read. Treat the snippets as illustrations and check the project's terms before copying them into a product. The sample SCPs on the [SCP mechanism page](/mechanisms/scp/) are different: they come from the aws-samples repository under MIT-0, which the library records for every entry. The shape is the same in all five. A selector picks a resource: a security group, an instance, a bucket, a Namespace, a template section. A condition tests one attribute. A result says deny, remove, block or fail. The differences are in what the engine can see and when. ## The five examples differ in what they read and what happens on a match Only the Custodian example changes the cloud. The others report or block. For the same intents in the clouds' own objects, see [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) and the [library](/library/), and for what each engine is, see [what is policy as code](/what-is-policy-as-code/). | Engine | Reads | On a match | Project license | | --- | --- | --- | --- | | [Open Policy Agent](/engines/open-policy-agent/) | Terraform plan JSON | Returns a deny message for the caller to act on | Apache 2.0 | | [Sentinel](/engines/hashicorp-sentinel/) | Terraform plan through imports | Pass or fail, at the enforcement level set in the product | None found | | [Cloud Custodian](/engines/cloud-custodian/) | Live S3 buckets | Runs the action, here removing statements | Apache 2.0 | | [Kyverno](/engines/kyverno/) | A Kubernetes Namespace | Blocks the request while failureAction is Enforce | Apache 2.0 | | [CloudFormation Guard](/engines/cloudformation-guard/) | A CloudFormation template | Reports a failing rule | Apache 2.0 | ## Related pages - [Compliance as code: rule bundles for AWS, Azure and Linux](/compliance-as-code/) - [Policy as code tools: five engines by the same eight fields](/policy-as-code-tools/) - [Open Policy Agent: fields, version and testing](/engines/open-policy-agent/) - [All of them](/guides/) ## Sources - Open Policy Agent: Terraform (Apache 2.0 project): https://www.openpolicyagent.org/docs/terraform (read 16 Nov 2026) - GitHub: OPA LICENSE: https://github.com/open-policy-agent/opa/blob/main/LICENSE (read 16 Nov 2026) - HashiCorp: define Sentinel policies in HCP Terraform: https://developer.hashicorp.com/terraform/cloud-docs/policy-enforcement/sentinel (read 16 Nov 2026) - GitHub: Cloud Custodian README: https://github.com/cloud-custodian/cloud-custodian (read 16 Nov 2026) - GitHub: Cloud Custodian LICENSE: https://raw.githubusercontent.com/cloud-custodian/cloud-custodian/main/LICENSE (read 16 Nov 2026) - Kyverno: validate rules: https://kyverno.io/docs/writing-policies/validate/ (read 16 Nov 2026) - GitHub: Kyverno LICENSE: https://raw.githubusercontent.com/kyverno/kyverno/main/LICENSE (read 16 Nov 2026) - GitHub: CloudFormation Guard README: https://raw.githubusercontent.com/aws-cloudformation/cloudformation-guard/main/README.md (read 16 Nov 2026) - GitHub: CloudFormation Guard LICENSE: https://raw.githubusercontent.com/aws-cloudformation/cloudformation-guard/main/LICENSE (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/policy-as-code-examples/ --- # Compliance as code packages control checks as deployable rule bundles, and most cloud bundles report rather than block Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Compliance as code means expressing control checks as files a machine can deploy and run. AWS Config conformance packs bundle Config rules and remediation actions, Azure Policy ships regulatory compliance initiatives, and the ComplianceAsCode project publishes host hardening content. Read on 16 Nov 2026, this page separates those bundles from guardrails that block. ## Four kinds of bundle exist, and they differ in where they run Cloud Custodian's documentation also has a Compliance as Code section; see [Cloud Custodian](/engines/cloud-custodian/) for the engine. | Bundle | What AWS, Microsoft or the project says | Runs against | | --- | --- | --- | | AWS Config conformance pack | A collection of AWS Config rules and remediation actions, deployed as a single entity in an account and Region or across an organization; authored as a YAML template | AWS resource configuration | | Azure Policy regulatory compliance initiative | Built-in initiative definitions listing controls and compliance domains by responsibility (Customer, Microsoft, Shared); the feature is marked Preview | Azure resources | | AWS Control Tower control groups | Controls are grouped by objective, framework, service and group, with a Frameworks category of industry-standard compliance frameworks | AWS accounts in an OU | | ComplianceAsCode content | Security automation content in SCAP, Bash, Ansible and other formats; repository topics include pci-dss, stig and security-hardening | Hosts and configurations | ## A bundle that reports does not stop the change that breaks it AWS says Config rules implement Control Tower's detective controls, which detect noncompliance after a resource exists and alert through a dashboard. Preventive controls are the ones implemented with service control policies and resource control policies. A conformance pack of Config rules is on the reporting side, though AWS also lets a pack carry remediation actions. This is the point where compliance as code meets the subject of this site. A pack that says an S3 bucket is public tells you after the fact. A guardrail such as [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) prevents the change. The split is explained in [preventive vs detective controls](/guides/preventive-vs-detective-controls/). ## Azure's initiative structure carries the compliance mapping inside the policy file Microsoft documents that an initiative definition is JSON that groups related policy definitions so they can be assigned as one item. It also contains policy groups, a property that belongs to the Regulatory Compliance (Preview) feature. The mapping from policy to control therefore lives in the initiative file itself. The built-in initiatives page lists items by standard, including a HITRUST/HIPAA initiative and PCI DSS v4. Microsoft says the controls map to the corresponding compliance standard, and that existing compliance standard initiatives are being updated to support Regulatory Compliance. Assigning one is the same act as assigning any policy: it can start in audit and later be enforced, as in [Azure Policy as a guardrail](/guides/azure-policy-as-a-guardrail/). ## Four questions decide whether a bundle fits a guardrail program The questions are this page's own checklist, built from the documented behavior above. - Does the bundle block, or only report? Read the effect or the control type, not the bundle name. - Does it cover accounts that are not enrolled? AWS says detective controls apply to enrolled accounts only. - Can it be tried first? Azure assignments can use DoNotEnforce, and Config rules can be deployed and read before remediation is attached. - Who owns an exception, and when does it end? See [guardrail exceptions and waivers](/guardrail-exceptions-and-waivers/). ## Passing the rules is evidence about configuration, not proof that a control is met A rule bundle checks the settings its authors encoded. A control statement in a standard may need an interview, a log review or a process. Which framework control each rule satisfies is a mapping job that this page does not cover. What a team can do with a bundle is keep the checks in version control, run them on a schedule and keep each result. For the engines that run such checks, see [policy as code tools](/policy-as-code-tools/), and for the layer of rules that block, see the [library](/library/) and the [method](/method/). ## Related pages - [Policy as code tools: five engines by the same eight fields](/policy-as-code-tools/) - [Open Policy Agent: fields, version and testing](/engines/open-policy-agent/) - [HashiCorp Sentinel: fields, version and testing](/engines/hashicorp-sentinel/) - [All of them](/guides/) ## Sources - AWS Config: conformance packs: https://docs.aws.amazon.com/config/latest/developerguide/conformance-packs.html (read 16 Nov 2026) - Microsoft: regulatory compliance in Azure Policy: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/regulatory-compliance (read 16 Nov 2026) - Microsoft: initiative definition structure: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/initiative-definition-structure (read 16 Nov 2026) - Microsoft: built-in initiatives: https://learn.microsoft.com/en-us/azure/governance/policy/samples/built-in-initiatives (read 16 Nov 2026) - AWS Control Tower: proactive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) - AWS Control Tower: detective controls: https://docs.aws.amazon.com/controltower/latest/controlreference/detective-controls.html (read 10 Nov 2026) - GitHub: ComplianceAsCode/content: https://github.com/ComplianceAsCode/content (read 16 Nov 2026) - Cloud Custodian: quick start: https://cloudcustodian.io/docs/quickstart/index.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/compliance-as-code/ --- # Five policy as code engines, listed alphabetically by the same fields, with no ranking and no score Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Five policy as code engines are described here by one set of fields: what each evaluates, where it runs, its language, license and current version, read on 16 Nov 2026. The order is alphabetical. This page gives no ranking and no score. Each engine has its own page with the testing step before enforcement. ## The five engines side by side The two dates marked without a year come from release pages that print the day and month only; the notes on each engine page explain how the year was read. | Engine | What it evaluates | Where it runs | Policy language | License | Current version | | --- | --- | --- | --- | --- | --- | | [Cloud Custodian](/engines/cloud-custodian/) | Live cloud resources | Cron, serverless event modes | YAML | Apache 2.0 | 0.9.53.0 (01 Oct, no year shown) | | [CloudFormation Guard](/engines/cloudformation-guard/) | JSON or YAML data, such as CloudFormation templates and Terraform plans | CLI, CI, pre-commit, CloudFormation Hooks | Guard DSL | Apache 2.0 | v3.2.1 (25 Aug 2026) | | [HashiCorp Sentinel](/engines/hashicorp-sentinel/) | Terraform plan, configuration and state through imports | Inside HCP Terraform and other HashiCorp products; local CLI | Sentinel language | None found on 16 Nov 2026 | 0.41.0 (date not found) | | [Kyverno](/engines/kyverno/) | Kubernetes resources | Cluster admission controller, CLI | YAML policies | Apache 2.0 | v1.20.0 (12 Nov) | | [Open Policy Agent](/engines/open-policy-agent/) | Any structured JSON input | Library, server, CLI; Kubernetes, CI/CD, Terraform | Rego | Apache 2.0 | v1.22.4 (12 Nov) | ## The testing step before enforcement is the field that varies most These are the engines' own mechanisms. The cloud-side equivalents are listed in [testing a guardrail before you apply it](/guides/testing-a-guardrail-before-you-apply-it/). | Engine | How a rule is tried before it blocks | | --- | --- | | Cloud Custodian | `custodian run --dryrun` lists matches and runs no actions | | CloudFormation Guard | `cfn-guard test` with JSON or YAML test files | | HashiCorp Sentinel | `sentinel test` with mock data; advisory enforcement level warns and does not block | | Kyverno | failureAction Audit reports to a policy report; the CLI tests policies | | Open Policy Agent | `opa test` unit tests; saved plans evaluated with `opa exec` | ## Each field was read from a named page, so a reader can recheck it Version numbers change often. The versions in the table were the latest listed on the day read, and the releases pages linked from each engine page show the current one. | Engine | Fields read from | | --- | --- | | Cloud Custodian | GitHub README and releases page, cloudcustodian.io, CNCF project page, quick start, LICENSE file | | CloudFormation Guard | GitHub README and releases page, AWS user guide pages on what Guard is and on testing, LICENSE file | | HashiCorp Sentinel | Sentinel documentation (concepts, language, testing, commands, downloads) and HCP Terraform policy enforcement pages | | Kyverno | GitHub README and releases page, kyverno.io validate and CLI pages, CNCF project page, LICENSE file | | Open Policy Agent | opa documentation (introduction, Terraform, policy testing), GitHub releases page, CNCF project page, LICENSE file | ## Maintainers range from foundations to a single vendor Open Policy Agent is a CNCF graduated project, and Cloud Custodian is a CNCF incubating project. The CNCF page for Kyverno lists graduation on 16 Mar 2026. CloudFormation Guard is maintained by AWS. Sentinel is a HashiCorp product, and the pages read give no public source repository: github.com/hashicorp/sentinel returned HTTP 404 on 16 Nov 2026. The OPA site also lists related projects, among them Gatekeeper for Kubernetes, Conftest and Regal. They are not described here because the fields above were read from each project's own site and repository. ## This index is separate from the platform ranking The engines are free-standing tools and are not in the ranking of 16 platforms that is built on [the method](/method/). The ranking asks how safely a platform puts a guardrail into production. This page does not. For the cloud-side objects these engines sit beside, see the [library](/library/) and [what is policy as code](/what-is-policy-as-code/). ## Related pages - [Open Policy Agent: fields, version and testing](/engines/open-policy-agent/) - [HashiCorp Sentinel: fields, version and testing](/engines/hashicorp-sentinel/) - [Cloud Custodian: fields, version and testing](/engines/cloud-custodian/) - [All of them](/guides/) ## Sources - Open Policy Agent: introduction: https://www.openpolicyagent.org/docs (read 16 Nov 2026) - GitHub: OPA releases: https://github.com/open-policy-agent/opa/releases (read 16 Nov 2026) - HashiCorp: install Sentinel: https://developer.hashicorp.com/sentinel/downloads (read 16 Nov 2026) - GitHub: Cloud Custodian releases: https://github.com/cloud-custodian/cloud-custodian/releases (read 16 Nov 2026) - GitHub: Kyverno releases: https://github.com/kyverno/kyverno/releases (read 16 Nov 2026) - GitHub: CloudFormation Guard releases: https://github.com/aws-cloudformation/cloudformation-guard/releases (read 16 Nov 2026) - CNCF: Kyverno: https://www.cncf.io/projects/kyverno/ (read 16 Nov 2026) - CNCF: Cloud Custodian: https://www.cncf.io/projects/cloud-custodian/ (read 16 Nov 2026) - CNCF: Open Policy Agent: https://www.cncf.io/projects/open-policy-agent-opa/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/policy-as-code-tools/ --- # Open Policy Agent evaluates any structured data with Rego policies and is a CNCF graduated project under Apache 2.0 Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Open Policy Agent (OPA) is a general-purpose policy engine that evaluates structured input, such as a Terraform plan in JSON, against policies written in Rego. It is Apache 2.0 licensed and a CNCF graduated project. The current release read on 16 Nov 2026 is v1.22.4, published 12 Nov 2026. ## Open Policy Agent described by eight fields, read 16 Nov 2026 | Field | Value | | --- | --- | | What it evaluates | Arbitrary structured data, such as JSON, against policies and data. OPA says decisions are not limited to allow or deny and can be structured output. | | Where it runs | Called by software that needs a decision. OPA documentation lists Kubernetes, CI/CD, Terraform, AWS CloudFormation Hooks, Envoy and Istio, and HTTP APIs as use cases. A CLI is included (`opa eval`, `opa exec`, `opa test`). | | Policy language | Rego, a declarative language | | Clouds and targets | Domain-agnostic. For clouds, the documented route is the JSON plan from `terraform show -json`; OPA reads the plan, not the cloud account. | | License | Apache License 2.0 (LICENSE file in the repository, read 16 Nov 2026) | | Maintainer | The Open Policy Agent project, a Cloud Native Computing Foundation project: accepted 29 Mar 2018, graduated 29 Jan 2021 (CNCF page, read 16 Nov 2026) | | Tested before enforcement | `opa test` runs rules named test_ against policy files; the Terraform guide applies unit tests to saved plans | | Current version | v1.22.4, released 12 Nov (the page shows no year; its asset timestamps read 2026-11-12). The release note calls it a fix release for a vulnerability in Go's HTTP/1 server | ## OPA separates the decision from the enforcement point, so the cloud action is blocked by something else OPA documentation says it "decouples policy decision-making from policy enforcement": software queries OPA with structured input and acts on the answer. OPA therefore does not itself deny an AWS API call or an Azure deployment. A pipeline step, an admission webhook or a Terraform workflow takes the answer and blocks. HashiCorp documents that HCP Terraform supports OPA as one of its policy frameworks, next to Sentinel and Terraform policy; a policy set can hold policies in only one framework. For the Terraform route the input is the JSON plan, so the guardrail is a pre-deploy check. A resource created by hand in the console never passes through it. ## A policy is tested with unit tests and then with a saved plan The OPA policy-testing page describes writing test files, such as example_test.rego with rules like test_get_anonymous_denied, and running `opa test . -v` to see PASS or FAIL lines. The Terraform guide builds on this: it saves a plan with `terraform show -json`, evaluates `data.terraform.module.deny` against it, and uses `opa exec` for the same decision. An audit-only mode that logs violations without blocking, for cloud accounts: none found on 16 Nov 2026 in the pages read. In practice the audit step is a CI job that reports and does not fail the build, which is a choice made in the pipeline, not a setting in OPA. ## The project states that plan-time data is incomplete OPA documentation lists what a Terraform plan may not contain: unknown values (computed attributes decided at apply time), dynamic blocks, and the results of Terraform functions not yet evaluated. It tells authors to confirm the needed information is in the plan JSON before writing a policy. This is the main limit for cloud guardrails written in Rego. A rule about an attribute that is computed at apply time cannot be checked in the plan. The cloud-side mechanisms, such as [service control policies](/mechanisms/scp/), see the final request instead. ## Rego sits upstream of the cloud's own guardrails, not in place of them A Rego rule that checks a plan for a public storage account complements an [Azure Policy deny effect](/mechanisms/azure-policy/) on the same setting: the first blocks the change in review, the second blocks it in the cloud. See [policy as code tools](/policy-as-code-tools/) for the other engines, [policy as code examples](/policy-as-code-examples/) for a real Rego rule, and the [library](/library/) for the cloud-side entries. ## Related pages - [HashiCorp Sentinel: fields, version and testing](/engines/hashicorp-sentinel/) - [Cloud Custodian: fields, version and testing](/engines/cloud-custodian/) - [Kyverno: fields, version and testing](/engines/kyverno/) - [All of them](/guides/) ## Sources - Open Policy Agent: introduction: https://www.openpolicyagent.org/docs (read 16 Nov 2026) - Open Policy Agent: Terraform: https://www.openpolicyagent.org/docs/terraform (read 16 Nov 2026) - Open Policy Agent: policy testing: https://www.openpolicyagent.org/docs/policy-testing (read 16 Nov 2026) - GitHub: OPA releases: https://github.com/open-policy-agent/opa/releases (read 16 Nov 2026) - GitHub: OPA LICENSE: https://github.com/open-policy-agent/opa/blob/main/LICENSE (read 16 Nov 2026) - CNCF: Open Policy Agent: https://www.cncf.io/projects/open-policy-agent-opa/ (read 10 Nov 2026) - HashiCorp: policy enforcement in HCP Terraform: https://developer.hashicorp.com/terraform/cloud-docs/policy-enforcement (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/engines/open-policy-agent/ --- # HashiCorp Sentinel enforces policies inside HashiCorp products, with three enforcement levels and a local test command Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. HashiCorp Sentinel is a policy-as-code framework embedded in HashiCorp products, including HCP Terraform, where its policies check Terraform plans. It has advisory, soft mandatory and hard mandatory enforcement levels and a sentinel test command. Version 0.41.0 is listed as latest on 16 Nov 2026; its license terms were not found. ## HashiCorp Sentinel described by eight fields, read 16 Nov 2026 | Field | Value | | --- | --- | | What it evaluates | Data that a host application supplies through imports. In HCP Terraform the imports are tfplan, tfconfig, tfstate and tfrun | | Where it runs | Inside Sentinel-enabled applications such as HCP Terraform; a local CLI is documented for development and testing | | Policy language | The Sentinel language: policies are single .sentinel files with a main rule | | Clouds and targets | Whatever Terraform plans. HashiCorp documents Terraform and Consul integrations | | License | None found on 16 Nov 2026 in the pages read. The documentation offers binary downloads, and github.com/hashicorp/sentinel returned HTTP 404 | | Maintainer | HashiCorp (developer.hashicorp.com documentation) | | Tested before enforcement | `sentinel test` runs cases in test//*.hcl or .json with mock data; the advisory enforcement level lets a policy fail with a warning | | Current version | 0.41.0, marked latest on the Sentinel downloads page; a release date: none found on 16 Nov 2026 | ## Sentinel runs where the host product runs it, and checks the plan before apply HashiCorp says Sentinel "is a policy as code tool that lets you control what users of HashiCorp products are allowed to do" and enforces policies proactively. In HCP Terraform, policies are rules that validate Terraform plans against security rules and best practices. HashiCorp gives the example of preventing production deployments to the wrong region. HCP Terraform applies policy sets to workspaces, to projects or globally, and checks the plan during each run. A policy set may contain policies from only one framework, but several sets with different frameworks can attach to one workspace. HashiCorp also publishes pre-written policies for common standards such as PCI DSS. A Sentinel policy is a single .sentinel file, and the language guide says there is no built-in way to merge several files. The result comes from a main rule: a boolean passes on true and fails on false. HashiCorp says the language can be learned in an hour; that is its claim, not a measurement. ## Enforcement levels and local tests are the two safety steps Sentinel makes the level a deployment setting, separate from the policy logic. Advisory, the default, lets a policy fail with a warning. Soft mandatory must pass unless a user with the right privilege overrides it. Hard mandatory must pass, and the only override is removing the policy. A new guardrail can therefore start as advisory and move up. The `sentinel test` command looks for cases under test//, where each HCL or JSON file is one case with mock data. HashiCorp recommends storing policies in version control and running these tests in CI. ## Reach is limited to what the HashiCorp product sees HashiCorp states that HCP Terraform does not support custom imports, so the data a policy can read is fixed to the four Terraform imports. A change made outside Terraform is never presented to the policy. The HCP Terraform Free edition includes one policy set of up to five policies, per HashiCorp's page; Standard and Premium editions add version control connections and API-created policy set versions. Check the current pricing page before relying on that. ## Sentinel and OPA can be used on the same Terraform workspace family Because HCP Terraform supports both frameworks, the choice is often between a HashiCorp-specific language and [Open Policy Agent](/engines/open-policy-agent/). Compare the fields on the [policy as code tools](/policy-as-code-tools/) page, see a real Sentinel rule on [policy as code examples](/policy-as-code-examples/), and read the [method](/method/) for how this site treats rollout safety. ## Related pages - [Cloud Custodian: fields, version and testing](/engines/cloud-custodian/) - [Kyverno: fields, version and testing](/engines/kyverno/) - [CloudFormation Guard: fields, version and testing](/engines/cloudformation-guard/) - [All of them](/guides/) ## Sources - HashiCorp: Sentinel documentation: https://developer.hashicorp.com/sentinel/docs (read 16 Nov 2026) - HashiCorp: Sentinel enforcement levels: https://developer.hashicorp.com/sentinel/docs/concepts/enforcement-levels (read 16 Nov 2026) - HashiCorp: testing Sentinel policies: https://developer.hashicorp.com/sentinel/docs/writing/testing (read 16 Nov 2026) - HashiCorp: sentinel test command: https://developer.hashicorp.com/sentinel/docs/commands/test (read 16 Nov 2026) - HashiCorp: install Sentinel: https://developer.hashicorp.com/sentinel/downloads (read 16 Nov 2026) - HashiCorp: Sentinel language: https://developer.hashicorp.com/sentinel/docs/language (read 16 Nov 2026) - HashiCorp: policy enforcement in HCP Terraform: https://developer.hashicorp.com/terraform/cloud-docs/policy-enforcement (read 16 Nov 2026) - HashiCorp: define Sentinel policies in HCP Terraform: https://developer.hashicorp.com/terraform/cloud-docs/policy-enforcement/sentinel (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/engines/hashicorp-sentinel/ --- # Cloud Custodian runs YAML policies against live AWS, Azure and Google Cloud resources, and can remediate after the fact Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cloud Custodian is an Apache 2.0 rules engine that reads YAML policies and acts on existing AWS, Azure and Google Cloud resources, either on a schedule or from cloud events. It is a CNCF incubating project. Release 0.9.53.0 is marked latest on the page read on 16 Nov 2026. Its dryrun mode lists matches without acting. ## Cloud Custodian described by eight fields, read 16 Nov 2026 | Field | Value | | --- | --- | | What it evaluates | Cloud resources (the README names EC2, ASG, Redshift, CosmosDB and Pub/Sub topics) through filters; the README also says it can run policies on infrastructure-as-code assets | | Where it runs | As a cron job on a server, locally, or serverless with cloud event sources: AWS CloudWatch Events and Config rules, Azure EventGrid, GCP audit log and Pub/Sub | | Policy language | YAML: each policy sets a resource type, filters, actions and a mode | | Clouds and targets | AWS, Azure and Google Cloud; Kubernetes, Tencent Cloud and OpenStack in beta per cloudcustodian.io | | License | Apache License 2.0 (LICENSE file, read 16 Nov 2026) | | Maintainer | The Cloud Custodian community, a CNCF incubating project: accepted 25 Jun 2020, incubating since 14 Sep 2022 | | Tested before enforcement | `custodian validate` checks the file and `custodian run --dryrun` shows which resources match, with no actions executed | | Current version | 0.9.53.0, marked Latest; the releases page shows 01 Oct 22:58 with no year, read 16 Nov 2026 | ## Custodian acts on resources that exist, so it is corrective more often than preventive The README says Custodian can enforce policies in real time by integrating with each provider's serverless features, or run as a cron job against large existing fleets. The Custodian documentation has separate mode pages for AWS, Azure and GCP. Its first example policy removes cross-account statements from S3 buckets. Its second runs on a CloudTrail RunInstances event and terminates any new EC2 instance whose volumes are not encrypted. That second policy is a guardrail in effect, but the instance is created first and removed after. This differs from a deny in a [service control policy](/mechanisms/scp/) or an Azure Policy deny effect, where the request fails. The README also lists Terraform support for policies on infrastructure-as-code, and cloudcustodian.io labels that feature Alpha. ## Dryrun is the test step, and it runs against live data The README shows three commands in order: validate the file, run with --dryrun and an output directory to see which resources match each policy without executing actions, then run. Because dryrun evaluates real resources, it shows the actual blast radius of a filter before an action such as terminate is allowed to run. A native unit test framework for policies, comparable to `opa test` or `cfn-guard test`: none found on 16 Nov 2026 in the pages read. ## Event-mode policies need cloud-side setup that the engine provisions The README says Custodian automatically provisions serverless functions and event sources. The quickstart policy names an IAM role, CloudCustodian-QuickStart, for its CloudTrail mode. That means the engine deploys code into the account it governs, and the identity it runs under becomes a privileged part of the account. Multi-account use is listed as supported. How a team limits that identity per account is a deployment decision; the pages read give the capability and not a recommended boundary. ## Pair it with preventive guardrails rather than treating it as one A common split is a preventive object for the rules that must never be broken, such as [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/), and Custodian for the rules that can be corrected after the fact. See [preventive vs detective controls](/guides/preventive-vs-detective-controls/) for that decision, [policy as code examples](/policy-as-code-examples/) for the YAML shape, and [policy as code tools](/policy-as-code-tools/) for the other engines. ## Related pages - [Kyverno: fields, version and testing](/engines/kyverno/) - [CloudFormation Guard: fields, version and testing](/engines/cloudformation-guard/) - [AWS SCP examples explained: what each pattern blocks and breaks](/guides/aws-scp-examples-explained/) - [All of them](/guides/) ## Sources - GitHub: cloud-custodian/cloud-custodian: https://github.com/cloud-custodian/cloud-custodian (read 16 Nov 2026) - GitHub: Cloud Custodian releases: https://github.com/cloud-custodian/cloud-custodian/releases (read 16 Nov 2026) - GitHub: Cloud Custodian LICENSE: https://raw.githubusercontent.com/cloud-custodian/cloud-custodian/main/LICENSE (read 16 Nov 2026) - Cloud Custodian home page: https://www.cloudcustodian.io/ (read 16 Nov 2026) - CNCF: Cloud Custodian: https://www.cncf.io/projects/cloud-custodian/ (read 16 Nov 2026) - Cloud Custodian: quick start: https://cloudcustodian.io/docs/quickstart/index.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/engines/cloud-custodian/ --- # Kyverno enforces policy on Kubernetes resources at admission, and its own README limits it to Kubernetes Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Kyverno is an Apache 2.0 policy engine for Kubernetes that validates, mutates and generates resources at admission and scans existing ones. The CNCF project page read on 16 Nov 2026 lists it as graduated on 16 Mar 2026. Release v1.20.0 is latest, published 12 Nov 2026. Its CLI tests policies before they reach a cluster. ## Kyverno described by eight fields, read 16 Nov 2026 | Field | Value | | --- | --- | | What it evaluates | Kubernetes resources: validation, mutation, generation and image verification, plus background scans of existing resources | | Where it runs | As an admission controller inside a cluster, and as a CLI for testing policies against resources before they are added to a cluster | | Policy language | YAML policies (ClusterPolicy under kyverno.io/v1); the documentation also lists newer policy types such as ValidatingPolicy and a guide to migrating to CEL policies | | Clouds and targets | Kubernetes clusters on any cloud. The README says Kyverno can only affect policies used by Kubernetes. Account-level guardrails for AWS, Azure or Google Cloud: none found on 16 Nov 2026 | | License | Apache License 2.0 (LICENSE file, read 16 Nov 2026) | | Maintainer | The Kyverno project, contributed by Nirmata. The CNCF page says graduated on 16 Mar 2026; the GitHub README read the same day still says incubating | | Tested before enforcement | failureAction Audit records violations in a policy report and allows the request; Enforce blocks. The Kyverno CLI tests policies against resources | | Current version | v1.20.0, marked Latest, released 12 Nov (the page shows no year; asset timestamps read 2026-11-11) | ## Kyverno governs what runs on clusters, not the cloud account that hosts them Kyverno's README says it is "only able to impact the policies used by Kubernetes". It does not replace Kubernetes RBAC, and it complements the built-in ValidatingAdmissionPolicies and MutatingAdmissionPolicies with reporting, exception management and periodic background scanning. Companion projects, such as Kyverno JSON for non-Kubernetes JSON payloads, Chainsaw for end-to-end testing and Policy Reporter for violation reporting, are separate and have their own release cycles. For this site that sets the boundary. A Kyverno policy can stop a privileged pod. It cannot stop an engineer attaching an IAM policy in AWS. The cloud-side entries in the [library](/library/) cover the second case. ## Audit mode and per-namespace overrides are the rollout path Kyverno documents failureAction with two values. Enforce blocks the creation or update of a non-compliant resource. Audit allows it and logs a violation in a PolicyReport or ClusterPolicyReport. A policy in Audit mode also reports on existing resources during periodic scans. failureActionOverrides lets one policy enforce in one namespace and audit in another, as in the documentation's example where the default namespace enforces and the test namespace audits. Kyverno also documents a behavior for existing violations: a newly created Enforce rule lets updates continue on resources that already violate it, unless validate.allowExistingViolations is set to false. The CLI page says it can validate and test policy behavior against resources before adding them to a cluster. ## The project states what it cannot do about Kubernetes itself The README says Kyverno does not protect against vulnerabilities in the Kubernetes API server or the underlying infrastructure, and that its enforcement may be bypassed if Kubernetes itself has a security flaw. It also notes that the field spec.validationFailureAction is deprecated in favor of a per-rule failureAction, which affects policies copied from older examples. ## Kyverno belongs in the engine comparison for clusters only Teams that run Kubernetes on AWS, Azure or Google Cloud often need both a cluster-level engine and the cloud's own guardrails. See [policy as code tools](/policy-as-code-tools/) for the field-by-field table, [policy as code examples](/policy-as-code-examples/) for a Kyverno ClusterPolicy from the project's documentation, and the [mechanisms](/mechanisms/scp/) that cover the account layer. ## Related pages - [CloudFormation Guard: fields, version and testing](/engines/cloudformation-guard/) - [AWS SCP examples explained: what each pattern blocks and breaks](/guides/aws-scp-examples-explained/) - [How SCP evaluation works: allow lists, deny lists, and the limits](/guides/how-scp-evaluation-works/) - [All of them](/guides/) ## Sources - GitHub: kyverno/kyverno: https://github.com/kyverno/kyverno (read 16 Nov 2026) - GitHub: Kyverno releases: https://github.com/kyverno/kyverno/releases (read 16 Nov 2026) - GitHub: Kyverno LICENSE: https://raw.githubusercontent.com/kyverno/kyverno/main/LICENSE (read 16 Nov 2026) - CNCF: Kyverno: https://www.cncf.io/projects/kyverno/ (read 16 Nov 2026) - Kyverno: validate rules: https://kyverno.io/docs/writing-policies/validate/ (read 16 Nov 2026) - Kyverno: CLI: https://kyverno.io/docs/kyverno-cli/ (read 16 Nov 2026) - Kyverno: how Kyverno works: https://kyverno.io/docs/introduction/how-kyverno-works/ (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/engines/kyverno/ --- # AWS CloudFormation Guard checks JSON and YAML against rules in its own language, and blocks only through CloudFormation Hooks Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS CloudFormation Guard (cfn-guard) is an Apache 2.0 command line tool that validates JSON or YAML, such as CloudFormation templates and Terraform plans, against rules in a purpose-built language. AWS says it gives no server-side enforcement without Hooks. Release v3.2.1 is latest as read on 16 Nov 2026, dated 25 Aug 2026. ## AWS CloudFormation Guard described by eight fields, read 16 Nov 2026 | Field | Value | | --- | --- | | What it evaluates | Structured JSON or YAML: CloudFormation templates and change sets, Terraform JSON plans and state, Kubernetes configurations, and AWS Config configuration items | | Where it runs | CLI in CI, as a pre-commit hook, as a GitHub Action or a Docker image; server-side enforcement uses CloudFormation Hooks | | Policy language | Guard DSL: clauses that evaluate to true or false, grouped into rules | | Clouds and targets | General-purpose, with an AWS Guard Rules Registry of rule sets for AWS resources | | License | Apache License 2.0 (LICENSE file, read 16 Nov 2026) | | Maintainer | AWS: the repository is in the aws-cloudformation GitHub organization and AWS hosts the user guide | | Tested before enforcement | Built-in unit tests: `cfn-guard test --rules-file --test-data ` with JSON or YAML test files | | Current version | v3.2.1, marked Latest, released 25 Aug 2026 per the GitHub releases page | ## Guard evaluates a file, so it is a shift-left check unless wired into Hooks AWS describes Guard as an open-source, general-purpose policy-as-code evaluation tool with a CLI. Rules can be split across several files, one per purpose such as S3, DynamoDB or access management, and run together against one input. The README names three domains: preventative governance (validate templates and Terraform plans before deployment), detective governance (validate AWS Config configuration items) and deployment safety (validate CloudFormation change sets, for example to prevent a table rename that forces a replacement). AWS states that Guard "doesn't provide server-side enforcement" and points to CloudFormation Hooks for server-side validation, where an operation can be blocked or warned. [Control Tower proactive controls](/aws-control-tower-guardrails/) are the managed form of that route. ## Guard has a unit test framework and a scoped test file format AWS documents test files as JSON or YAML with multiple inputs and the expected outcome for each rule, and recommends starting with empty inputs before adding cases. The README notes that Guard 3.0 changed intrinsic function handling, so tests written for earlier versions may need review, and that Guard 2.0 and later are backward incompatible with 1.0 rules. The README says the validate command accepts a --structured flag that emits JSON or YAML output a pipeline can parse. AWS says Hooks can block or warn an operation, so a warn-first rollout is possible; the Hook settings that choose between them are outside the pages read. ## The project lists two things Guard does not do AWS says Guard does not validate CloudFormation templates for valid syntax or allowed property values, and recommends cfn-lint for that. It also gives no server-side enforcement by itself. A rule that passes Guard can still fail at deploy time, and a template that is never passed to Guard is never checked. The Control Tower documentation says its proactive controls, which are built on CloudFormation Hooks, evaluate the strings passed to the hook and cannot resolve secure strings or secrets. A rule about a value kept in a secret therefore needs a different check. ## Guard rules cover infrastructure code, so cloud-account guardrails sit beside them A Guard rule on S3 encryption in a template and the [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/) SCP address nearby risks at different points. See [policy as code examples](/policy-as-code-examples/) for the Guard syntax, [policy as code tools](/policy-as-code-tools/) for the other engines and [compliance as code](/compliance-as-code/) for the Config-based form. ## Related pages - [AWS SCP examples explained: what each pattern blocks and breaks](/guides/aws-scp-examples-explained/) - [How SCP evaluation works: allow lists, deny lists, and the limits](/guides/how-scp-evaluation-works/) - [Resource control policies: what RCPs are and how they differ from SCPs](/guides/resource-control-policies/) - [All of them](/guides/) ## Sources - GitHub: aws-cloudformation/cloudformation-guard: https://github.com/aws-cloudformation/cloudformation-guard (read 16 Nov 2026) - GitHub: CloudFormation Guard README: https://raw.githubusercontent.com/aws-cloudformation/cloudformation-guard/main/README.md (read 16 Nov 2026) - GitHub: CloudFormation Guard releases: https://github.com/aws-cloudformation/cloudformation-guard/releases (read 16 Nov 2026) - GitHub: CloudFormation Guard LICENSE: https://raw.githubusercontent.com/aws-cloudformation/cloudformation-guard/main/LICENSE (read 16 Nov 2026) - AWS: what is CloudFormation Guard: https://docs.aws.amazon.com/cfn-guard/latest/ug/what-is-guard.html (read 16 Nov 2026) - AWS: testing Guard rules: https://docs.aws.amazon.com/cfn-guard/latest/ug/testing-rules.html (read 16 Nov 2026) - AWS Control Tower: proactive controls: https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/engines/cloudformation-guard/ --- # 31 terms, each with a sourced example Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Plain definitions of the terms used in guardrail work, among them SCP, RCP, permissions boundary, declarative policy, deny assignment, organization policy constraint, perimeter and dry run, each with a sourced example and the reason it matters when approving a policy. - [Azure Policy](/glossary/azure-policy/): The Azure service that evaluates resources against business rules written as policy definitions, reports compliance, and can block, modify or remediate resources through an effect. - [Boolean constraint](/glossary/boolean-constraint/): A Google Cloud constraint that is either enforced or not enforced, with no list of values. - [Custom constraint](/glossary/custom-constraint/): A Google Cloud constraint written by your organization, in a YAML file that names the resources, methods, conditions and actions it applies to. - [Data perimeter](/glossary/data-perimeter/): A set of preventive guardrails that keeps only trusted identities reaching trusted resources from expected networks. - [Declarative policy](/glossary/declarative-policy/): An AWS Organizations policy type that states the desired configuration of a service across an organization and keeps it in place, enforced in the service control plane instead of at the authorization layer. - [Delegated administrator](/glossary/delegated-administrator/): A member account that an AWS service or Organizations feature designates to run that service for the organization, so the management account is used less. - [Deny assignment](/glossary/deny-assignment/): An Azure RBAC object that attaches a set of deny actions to a user, group or service principal at a scope. - [Detective control](/glossary/detective-control/): A control that finds noncompliance after a resource exists and reports it through a dashboard or alert, without stopping the change. - [Dry-run mode](/glossary/dry-run-mode/): A Google Cloud mode in which an organization policy is created and enforced like any other, but violations are written to audit logs and the violating actions are not denied. - [Enforcement mode](/glossary/enforcement-mode/): A property of an Azure Policy assignment that controls whether the effect is applied. - [Explicit deny](/glossary/explicit-deny/): A policy statement with a Deny effect that matches a request. - [Guardrail](/glossary/guardrail/): A policy that sets limits on what identities or resources in a cloud estate can do, without granting any permission itself. - [IAM deny policy](/glossary/iam-deny-policy/): A Google Cloud IAM policy of deny rules attached to a project, folder or organization. - [Implicit deny](/glossary/implicit-deny/): The default outcome when no policy allows an action. - [List constraint](/glossary/list-constraint/): A Google Cloud constraint whose policy holds a list of allowed or denied values, such as a list of permitted domains or Regions. - [Management account](/glossary/management-account/): The AWS account that created the organization and manages it. - [Management group](/glossary/management-group/): An Azure governance scope above subscriptions. - [Organization policy](/glossary/organization-policy/): A Google Cloud policy that sets one constraint on an organization, folder or project and applies to its descendants. - [Organizational unit](/glossary/organizational-unit/): A container for accounts inside an AWS organization. - [Permissions boundary](/glossary/permissions-boundary/): An AWS IAM feature that sets the maximum permissions an identity-based policy can grant to one IAM user or role. - [Policy assignment](/glossary/policy-assignment/): The Azure object that applies a policy definition or initiative to a scope and supplies its parameters. - [Policy effect](/glossary/policy-effect/): The single action an Azure Policy definition takes when its rule matches: deny, audit, modify, deployIfNotExists and others. - [Policy exemption](/glossary/policy-exemption/): An Azure Policy object that exempts a resource or resource hierarchy from evaluation of an assignment, with a category and an optional expiry. - [Policy initiative](/glossary/policy-initiative/): A group of Azure Policy definitions assigned together as one unit, also called a policySet. - [Preventive control](/glossary/preventive-control/): A control that disallows an action that would lead to a policy violation, so the request fails and the noncompliant state never exists. - [Principal access boundary](/glossary/principal-access-boundary/): A Google Cloud IAM policy attached to a set of principals that defines which resources those principals are eligible to access. - [Proactive control](/glossary/proactive-control/): A control that checks a resource before it is deployed, usually inside an infrastructure-as-code pipeline, and fails the deployment if the resource would be noncompliant. - [Resource control policy](/glossary/resource-control-policy/): An AWS Organizations policy that sets the maximum permissions on resources in member accounts, whoever the caller is. - [Resource lock](/glossary/resource-lock/): An Azure management lock on a subscription, resource group or resource that blocks deletion or modification regardless of the user's permissions. - [Service control policy](/glossary/service-control-policy/): An AWS Organizations policy that sets the maximum permissions for IAM users and roles in member accounts. - [VPC Service Controls perimeter](/glossary/vpc-service-controls-perimeter/): A Google Cloud boundary around projects and supported services that controls data egress and ingress across it, independent of IAM. ## Related pages - [News](/news/) - [Preventive guardrail mechanisms in AWS, Azure and Google Cloud](/mechanisms/) - [Platforms by how safely they put a guardrail into production](/platforms/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/ --- # Guardrail Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A policy that sets limits on what identities or resources in a cloud estate can do, without granting any permission itself. Guardrails are written once at a high level, such as an organization root or management group, and apply to everything beneath it. ## An example, sourced AWS Organizations documentation describes a service control policy as defining "a permission guardrail, or sets limits" on the actions IAM users and roles can perform. The page adds that no permissions are granted by an SCP: administrators must still attach identity-based or resource-based policies. The guardrail only narrows what those other policies can allow in member accounts. The library applies the term across all three clouds. ## Why it matters here The word covers several mechanisms across three clouds, and each blocks differently. Knowing that a guardrail grants nothing tells you where to look when a request fails: the cap is lower than the grant, and the cap is what to change. ## Related pages - [Preventive control: meaning and an example](/glossary/preventive-control/) - [Detective control: meaning and an example](/glossary/detective-control/) - [Proactive control: meaning and an example](/glossary/proactive-control/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/guardrail/ --- # Preventive control Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A control that disallows an action that would lead to a policy violation, so the request fails and the noncompliant state never exists. Preventive controls are implemented as deny-type policies attached at a scope, and they act at request time. ## An example, sourced AWS Control Tower documents a preventive control as one that "ensures that your accounts maintain compliance, because it disallows actions that lead to policy violations." Its status is either enforced or not enabled. The page says these controls are implemented using service control policies or resource control policies, both part of AWS Organizations. The same page notes that preventive controls apply to all member accounts under a registered OU, enrolled and unenrolled. ## Why it matters here A preventive control is the only kind that holds when nobody is watching. It also fails real work, so each one needs a tested exception path and a named owner before it is attached. Each one also needs an exception path and a named owner, so plan both before attaching it. ## Related pages - [Detective control: meaning and an example](/glossary/detective-control/) - [Proactive control: meaning and an example](/glossary/proactive-control/) - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/controltower/latest/controlreference/preventive-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/preventive-control/ --- # Detective control Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A control that finds noncompliance after a resource exists and reports it through a dashboard or alert, without stopping the change. Detective controls show state, so they suit rules that are still being learned and the gaps a preventive control leaves. ## An example, sourced AWS Control Tower documents that "a detective control detects noncompliance of resources within your accounts" and alerts through the dashboard. Its status is clear, in violation or not enabled. The page says detective controls are implemented using AWS Config rules, and that their names typically begin with the word Detect. The page says most of the strongly recommended controls, and many of the elective controls, that AWS Control Tower owns are detective. ## Why it matters here Detective controls are how you test a rule on real resources before you turn it into a block, and how you find what a block did not cover, such as the management account. They also show what a block missed, which makes them a companion to any preventive rule you ship. ## Related pages - [Proactive control: meaning and an example](/glossary/proactive-control/) - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [Resource control policy: meaning and an example](/glossary/resource-control-policy/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/controltower/latest/controlreference/detective-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/detective-control/ --- # Proactive control Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A control that checks a resource before it is deployed, usually inside an infrastructure-as-code pipeline, and fails the deployment if the resource would be noncompliant. It acts earlier than a preventive control but only on the paths it hooks. It acts earlier. ## An example, sourced AWS Control Tower documents proactive controls as checks that run when resources are created or updated through CloudFormation stack operations, implemented as preCreate and preUpdate hook handlers. The page warns the controls may not affect requests made directly through the console, the AWS APIs, the SDKs or other infrastructure-as-code tools. The page also notes that secure strings and secrets are not resolved before they reach the hook, so a control cannot evaluate them. ## Why it matters here A proactive control gives fast feedback to engineers who deploy through CloudFormation, and it leaves every other path open. Count it as a pipeline check, not as enforcement. It is an early signal for engineers using that tool and nothing more, so pair it with a preventive control for rules that must hold on every path. ## Related pages - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [Resource control policy: meaning and an example](/glossary/resource-control-policy/) - [Declarative policy: meaning and an example](/glossary/declarative-policy/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/controltower/latest/controlreference/proactive-controls.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/proactive-control/ --- # Service control policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An AWS Organizations policy that sets the maximum permissions for IAM users and roles in member accounts. It never grants access. It is attached to the root, an OU or an account, and the effective permissions are the intersection of the SCPs with the identity and resource policies. ## An example, sourced AWS documents that for a permission to be allowed in an account, an explicit Allow must exist at every level from the root through each OU to the account. When SCPs are enabled, AWS attaches a managed policy named FullAWSAccess to every root, OU and account. A Deny at any level on that path blocks the permission below it. A Deny-only SCP at the root, with no Allow, blocks every service. ## Why it matters here SCPs are the main preventive mechanism on AWS. Their evaluation rule explains most surprises, including the account where a new deny seems to do nothing because the principal sits in the management account. They also leave resources and outside principals alone, so decide early which risks need a resource control policy as well. ## Related pages - [Resource control policy: meaning and an example](/glossary/resource-control-policy/) - [Declarative policy: meaning and an example](/glossary/declarative-policy/) - [Permissions boundary: meaning and an example](/glossary/permissions-boundary/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_evaluation.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/service-control-policy/ --- # Resource control policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An AWS Organizations policy that sets the maximum permissions on resources in member accounts, whoever the caller is. It never grants access. Where an SCP caps your own principals, an RCP also reaches principals outside the organization. Management account resources stay outside it. ## An example, sourced AWS documents the choice in one note: use an SCP to limit the permissions of IAM principals within the organization member accounts, and use an RCP to restrict principals external to the organization that request access to resources in those accounts. The page lists the services that support RCPs, including Amazon S3, AWS STS and AWS KMS. AWS adds that RCPs do not apply to resources in the management account. ## Why it matters here RCPs close the gap SCPs leave for outside callers, and they have no simulator support, so they need a test account and CloudTrail review before they reach a wider scope. Review partner and vendor access first. Anyone outside your organization who relies on an unencrypted connection or an old TLS version is affected by the library entries. ## Related pages - [Declarative policy: meaning and an example](/glossary/declarative-policy/) - [Permissions boundary: meaning and an example](/glossary/permissions-boundary/) - [Explicit deny: meaning and an example](/glossary/explicit-deny/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/resource-control-policy/ --- # Declarative policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An AWS Organizations policy type that states the desired configuration of a service across an organization and keeps it in place, enforced in the service control plane instead of at the authorization layer. It holds even when the service adds features or APIs. ## An example, sourced AWS documents that declarative policies "centrally declare and enforce your desired configuration for a given AWS service at scale across an organization." The page gives blocking public internet access to Amazon VPC resources as an example and says they are enforced in the service control plane, unlike SCPs and RCPs, which regulate access to APIs. AWS lists the supported EC2 attributes, including VPC Block Public Access, Serial Console Access and Instance Metadata. ## Why it matters here A declarative policy does not depend on listing every API that could change a setting, which is where a deny-list SCP goes stale. The library covers VPC Block Public Access, snapshot sharing and IMDSv2 this way. Account administrators also cannot change an enforced value, which suits baseline settings that no team should override. ## Related pages - [Permissions boundary: meaning and an example](/glossary/permissions-boundary/) - [Explicit deny: meaning and an example](/glossary/explicit-deny/) - [Implicit deny: meaning and an example](/glossary/implicit-deny/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/declarative-policy/ --- # Permissions boundary Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An AWS IAM feature that sets the maximum permissions an identity-based policy can grant to one IAM user or role. The entity can do only what both the boundary and its identity policies allow. A boundary alone grants nothing. Think of it as a ceiling. ## An example, sourced AWS documents a user named Shirley who should manage only Amazon S3, Amazon CloudWatch and Amazon EC2. A boundary allows s3:*, cloudwatch:* and ec2:*. When a second policy allowing iam:CreateUser is attached, creating a user fails, because the boundary does not allow the action. Shirley can never act in IAM through that user. AWS also documents using a boundary to delegate user creation, and warns that a NotPrincipal Deny in a resource policy denies every principal carrying a boundary. ## Why it matters here Boundaries make it safe to let someone create identities, because the new identities cannot exceed the boundary. They apply per identity, so they do not replace an SCP baseline. The same cap can also be set on roles, which makes them fit for pipelines and automation that create other roles. Test each one before use. ## Related pages - [Explicit deny: meaning and an example](/glossary/explicit-deny/) - [Implicit deny: meaning and an example](/glossary/implicit-deny/) - [Organizational unit: meaning and an example](/glossary/organizational-unit/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/permissions-boundary/ --- # Explicit deny Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A policy statement with a Deny effect that matches a request. In AWS, an explicit deny in any applicable policy overrides every allow, so the request fails. A deny is the one policy statement no other policy can override. It cannot be overridden. ## An example, sourced AWS documents that within an account the permissions for an entity can be affected by identity-based policies, resource-based policies, permissions boundaries, Organizations SCPs or session policies, and states: "If any one of these policy types explicitly denies access for an operation, then the request is denied." Deny in an identity policy or a boundary overrides the allow. The same page says an explicit deny in an identity-based policy or a boundary overrides the allow. ## Why it matters here Most guardrails in the library are explicit denies, because a deny holds wherever it is attached and survives later grants. Look for the deny first when you debug an access failure. If a request fails and the identity policy allows it, a deny somewhere in the stack is the first thing to rule out. The simulator names the matching policy. ## Related pages - [Implicit deny: meaning and an example](/glossary/implicit-deny/) - [Organizational unit: meaning and an example](/glossary/organizational-unit/) - [Management account: meaning and an example](/glossary/management-account/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/explicit-deny/ --- # Implicit deny Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The default outcome when no policy allows an action. In AWS, anything not explicitly allowed is denied. Allow-list SCPs depend on it: an action missing from the allow list at any level is blocked without any deny statement being written. ## An example, sourced AWS documents that SCP evaluation "follows a deny-by-default model, meaning that any permissions not explicitly allowed in the SCPs are denied." If an allow statement is missing at the root, at the Production OU or at the account, access is denied. The page also notes that relying solely on allow statements can lead to unintended access because overlapping allows can override narrower ones. ## Why it matters here Implicit deny is why removing FullAWSAccess without a replacement makes every action in member accounts fail. Know which of your SCPs rely on it. A new AWS service is blocked under such an SCP until someone names it, which makes the allow-list pattern safer but more work. Choose it deliberately. ## Related pages - [Organizational unit: meaning and an example](/glossary/organizational-unit/) - [Management account: meaning and an example](/glossary/management-account/) - [Delegated administrator: meaning and an example](/glossary/delegated-administrator/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps_evaluation.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/implicit-deny/ --- # Organizational unit Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A container for accounts inside an AWS organization. Policies attached to an OU apply to the accounts and child OUs beneath it. OUs give you a scope smaller than the whole organization for testing a guardrail. Each OU can itself contain further OUs. ## An example, sourced The AWS Organizations quotas page documents that OUs can nest five levels deep under a root, and that each root, OU and account can have at most 10 service control policies attached directly. Inherited policies do not count against that limit. AWS states all policy limits are hard limits. AWS also states in the SCP guidance that you should create an OU and move accounts into it one at a time or in small numbers, so a mistake locks out few users. ## Why it matters here OUs are the staging area for AWS rollouts. AWS recommends moving accounts into a test OU one at a time before attaching an SCP at the root, so the OU layout is part of the guardrail design. Design the OU tree so a pilot group exists at each level before any guardrail needs one. ## Related pages - [Management account: meaning and an example](/glossary/management-account/) - [Delegated administrator: meaning and an example](/glossary/delegated-administrator/) - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/organizational-unit/ --- # Management account Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The AWS account that created the organization and manages it. Service control policies and resource control policies do not restrict its users, roles or resources, so guardrails that must cover it need another mechanism or tight access to the account itself. ## An example, sourced AWS documents that "SCPs don't affect users or roles in the management account. They affect only the member accounts in your organization." The same page lists the tasks SCPs cannot restrict, starting with any action performed by the management account, and any action done with permissions attached to a service-linked role. The RCP page states the same for resources: RCPs have no effect on resources in the management account, and they do not cover service-linked roles. ## Why it matters here An SCP test run from the management account proves nothing. Keep workloads out of the management account, because no SCP or RCP applies there. A guardrail that must cover those resources needs another mechanism. Keep the number of identities and workloads in the account small for the same reason, and review it often. ## Related pages - [Delegated administrator: meaning and an example](/glossary/delegated-administrator/) - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [Azure Policy: meaning and an example](/glossary/azure-policy/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/management-account/ --- # Delegated administrator Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A member account that an AWS service or Organizations feature designates to run that service for the organization, so the management account is used less. A delegated administrator is still a member account and still subject to the organization guardrails. ## An example, sourced AWS documents that SCPs "apply to member accounts that are designated as delegated administrators," which means a deny attached to an OU also reaches the account running a security or governance service. The RCP page states the same for resource control policies. The management account remains outside both. The page points to its best practices for the management account for more information, and the RCP page makes the same statement for resource control policies. ## Why it matters here An SCP that denies actions the delegated administrator needs can stop a security service from working. Check the delegated accounts before widening a deny. Test any new deny against the delegated accounts first, because a security service that cannot call its own APIs stops reporting without a visible error. ## Related pages - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [Azure Policy: meaning and an example](/glossary/azure-policy/) - [Policy effect: meaning and an example](/glossary/policy-effect/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/delegated-administrator/ --- # Data perimeter Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A set of preventive guardrails that keeps only trusted identities reaching trusted resources from expected networks. The aim is one always-on boundary around identities, resources and networks. Guardrails like these are always on, and they apply whatever teams build, so teams cannot opt out. ## An example, sourced The AWS whitepaper "Building a Data Perimeter on AWS" says many organizations want perimeter controls to help protect against unintended access and configuration errors through always-on permissions guardrails. It outlines best practices and services for creating a perimeter around identities, resources and networks in AWS, and was first published on June 13, 2023. The library's RCP entries describe the resource side of it. ## Why it matters here The perimeter is the main reason RCPs exist. A perimeter that uses SCPs alone covers your principals and leaves outside callers to resource policies. A perimeter turns many separate guardrails into one design goal. Name the objective each guardrail serves, whether identity, resource or network, so gaps become visible. ## Related pages - [Azure Policy: meaning and an example](/glossary/azure-policy/) - [Policy effect: meaning and an example](/glossary/policy-effect/) - [Enforcement mode: meaning and an example](/glossary/enforcement-mode/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/whitepapers/latest/building-a-data-perimeter-on-aws/building-a-data-perimeter-on-aws.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/data-perimeter/ --- # Azure Policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The Azure service that evaluates resources against business rules written as policy definitions, reports compliance, and can block, modify or remediate resources through an effect. Definitions are assigned to a scope such as a management group or subscription. Initiatives group several definitions into one assignment. ## An example, sourced Microsoft documents that Azure Policy "evaluates resources and actions in Azure by comparing the properties of those resources to business rules," described in JSON as policy definitions. Several definitions can be grouped into an initiative. The page lists evaluation triggers, including the standard compliance evaluation cycle that occurs once every 24 hours. Microsoft also lists built-in definitions such as Allowed Locations, whose Deny effect restricts the available locations for new resources. ## Why it matters here Azure Policy is the closest Azure analog of an SCP, but it acts on resource properties and not on identities. Nothing is blocked unless the effect is deny or a similar enforcing effect. Policy is where most Azure guardrails live. Check effect and mode before assuming a definition blocks. ## Related pages - [Policy effect: meaning and an example](/glossary/policy-effect/) - [Enforcement mode: meaning and an example](/glossary/enforcement-mode/) - [Policy assignment: meaning and an example](/glossary/policy-assignment/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/azure-policy/ --- # Policy effect Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The single action an Azure Policy definition takes when its rule matches: deny, audit, modify, deployIfNotExists and others. The effect decides whether a request is blocked, logged, changed or followed by a deployment. Each definition carries exactly one effect, and parameters can let assignments choose among several. ## An example, sourced Microsoft documents that each policy definition "has a single effect in its policyRule." The listed effects include addToNetworkGroup, append, audit, auditIfNotExists, deny, denyAction, deployIfNotExists, disabled, manual, modify and mutate. The order of evaluation puts disabled first, then append and modify, then deny, then audit, then denyAction last. Microsoft also notes that audit, deny, and either modify or append are often interchangeable, while manual is not, and disabled can replace any effect. ## Why it matters here Because modify runs before deny, a policy can change a request so that a deny never fires. Define the effect order for each definition you stack. The effect is the setting you change when you move from reporting to blocking. Treat it as a parameter, not as a redesign. ## Related pages - [Enforcement mode: meaning and an example](/glossary/enforcement-mode/) - [Policy assignment: meaning and an example](/glossary/policy-assignment/) - [Policy exemption: meaning and an example](/glossary/policy-exemption/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-basics (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/policy-effect/ --- # Enforcement mode Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A property of an Azure Policy assignment that controls whether the effect is applied. In DoNotEnforce the assignment is evaluated and compliance is reported, but the effect is not enforced on create or update and no activity log entry is written for it. ## An example, sourced Microsoft documents the enforcementMode property with the values Default, DoNotEnforce and Enroll. DoNotEnforce is described as useful when you want to evaluate the assignment before enforcement, for testing a new assignment or validating policy logic. Remediation tasks for deployIfNotExists policies can still be started when the mode is DoNotEnforce. The Enroll value applies the effect only to resources enrolled in the assignment, which suits adding subscopes in batches. ## Why it matters here DoNotEnforce is Azure's built-in report-only phase. Run it for a full evaluation cycle before switching an assignment to Default. It is the lowest-risk way to see the effect of an Azure guardrail on real resources. A policy in DoNotEnforce is easy to forget, so give each such assignment a review date and an owner. ## Related pages - [Policy assignment: meaning and an example](/glossary/policy-assignment/) - [Policy exemption: meaning and an example](/glossary/policy-exemption/) - [Policy initiative: meaning and an example](/glossary/policy-initiative/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/enforcement-mode/ --- # Policy assignment Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The Azure object that applies a policy definition or initiative to a scope and supplies its parameters. The assignment decides which resources are evaluated, the enforcement mode, the excluded scopes, and any overrides of the effect. It can also exclude subscopes. ## An example, sourced Microsoft documents that "policy assignments define which resources are evaluated by a policy definition or initiative," and that the assignment can set parameter values at assignment time. The page describes notScopes for exclusions and says one assignment can hold up to 10 overrides, each covering up to 50 policy definition reference IDs. Microsoft also documents that exclusions set with notScopes can be added or updated after the assignment is created. ## Why it matters here Most Azure guardrail behavior lives in the assignment, not the definition. Review the scope, mode and exclusions of an assignment before reading its rule. A definition can look right while its assignment is wrong. Read both when a guardrail fails to block, and check the assignment for the scope and mode first. ## Related pages - [Policy exemption: meaning and an example](/glossary/policy-exemption/) - [Policy initiative: meaning and an example](/glossary/policy-initiative/) - [Management group: meaning and an example](/glossary/management-group/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/policy-assignment/ --- # Policy exemption Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An Azure Policy object that exempts a resource or resource hierarchy from evaluation of an assignment, with a category and an optional expiry. The exempted resource stays tracked, unlike an excluded scope. Exemptions are meant for time-bound or specific cases and appear in compliance reports. ## An example, sourced Microsoft documents two categories: Mitigated, where the policy intent is met another way, and Waiver, where non-compliance is temporarily accepted. The optional expiresOn property sets when the exemption stops being honored. The page states the exemption object is not deleted at that date: it is preserved for record-keeping but no longer honored. The page also lets an exemption apply to only some definitions inside an initiative, through policy definition reference IDs. ## Why it matters here A dated Waiver is the exception path to plan before enforcement. It records who accepted the risk and until when, which an exclusion does not. A dated waiver turns an informal exception into a record with an owner. Set expiresOn on every one, because an exemption without an expiry stays honored indefinitely. ## Related pages - [Policy initiative: meaning and an example](/glossary/policy-initiative/) - [Management group: meaning and an example](/glossary/management-group/) - [Deny assignment: meaning and an example](/glossary/deny-assignment/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/concepts/exemption-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/policy-exemption/ --- # Policy initiative Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A group of Azure Policy definitions assigned together as one unit, also called a policySet. Initiatives let a baseline be assigned, tracked and exempted as a whole, or in part by definition reference ID. Many guardrails can ride on one assignment. ## An example, sourced Microsoft documents that "to simplify management, several business rules can be grouped together to form a policy initiative, also referred to as a policySet." The definition or initiative is then assigned to a scope of resources that Azure supports, such as management groups, subscriptions, resource groups or individual resources, and subscopes can be excluded. Microsoft also states that Azure Policy evaluates when a scope gets a new assignment, and when an assigned initiative is updated. ## Why it matters here A baseline of guardrails is easier to roll out as one initiative with one enforcement mode. An exemption can still target a single definition inside it. An initiative also lets the effect and the exemption category vary by definition inside it, which makes it practical to share a baseline across teams. ## Related pages - [Management group: meaning and an example](/glossary/management-group/) - [Deny assignment: meaning and an example](/glossary/deny-assignment/) - [Resource lock: meaning and an example](/glossary/resource-lock/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/policy-initiative/ --- # Management group Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An Azure governance scope above subscriptions. Conditions applied to a management group cascade by inheritance to all its subscriptions, so one policy assignment can cover many subscriptions in a tree. A tree of groups can mirror an organization's structure, and the root group sits above all of them. ## An example, sourced Microsoft documents that management groups provide a governance scope above subscriptions and that governance conditions applied there "cascade by inheritance to all associated subscriptions." A single directory supports 10,000 management groups, and a tree can be up to six levels deep. The example applies a policy that limits VM locations to West US on a management group named Corp. The tree is rooted in a root management group for each directory. ## Why it matters here Management groups are Azure's OU. Assign guardrails at a lower group first, as you would use a test OU on AWS. Management groups let a single assignment reach many subscriptions at once. Assign guardrails at a lower group first and move up, because everything below inherits the assignment. ## Related pages - [Deny assignment: meaning and an example](/glossary/deny-assignment/) - [Resource lock: meaning and an example](/glossary/resource-lock/) - [Organization policy: meaning and an example](/glossary/organization-policy/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/management-group/ --- # Deny assignment Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An Azure RBAC object that attaches a set of deny actions to a user, group or service principal at a scope. It blocks actions even when a role assignment grants them. In generally available scenarios, Azure creates and manages them. ## An example, sourced Microsoft documents that a deny assignment "attaches a set of deny actions to a user, group, or service principal at a particular scope for the purpose of denying access." Customer-managed deny assignments are in private preview, and for generally available scenarios Azure creates them, for example when deploying a deployment stack with deny settings. The page lists deny assignment properties that include excluded principals and excluded control plane and data plane actions, so a deny can carry its own exceptions. ## Why it matters here You cannot write a deny assignment directly in general availability. The route to one is a deployment stack, so plan it in your infrastructure code. Because only Azure creates these in general availability, find the deployment stack that owns one before you try to work around it. Its deny settings, not a hand-written rule, define the block. ## Related pages - [Resource lock: meaning and an example](/glossary/resource-lock/) - [Organization policy: meaning and an example](/glossary/organization-policy/) - [List constraint: meaning and an example](/glossary/list-constraint/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/deny-assignment/ --- # Resource lock Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. An Azure management lock on a subscription, resource group or resource that blocks deletion or modification regardless of the user's permissions. The two levels are CanNotDelete and ReadOnly. Locks protect management-plane operations only. Locks are applied separately from policy and role assignments. ## An example, sourced Microsoft documents that you can lock a subscription, resource group or resource "to protect them from accidental user deletions and modifications," and that the lock overrides any user permissions. CanNotDelete lets authorized users read and modify but not delete. ReadOnly means read only, similar to the Reader role. A ReadOnly lock on an SQL logical server still allows data operations. ## Why it matters here Locks are a coarse backstop for resources that must not be deleted. They can block operations that do not look like changes, so test one on a non-production copy. Locks sit outside policy, so a deny assignment or policy review will not show them. Keep an inventory of locks on production resources, and note who may create or remove them. ## Related pages - [Organization policy: meaning and an example](/glossary/organization-policy/) - [List constraint: meaning and an example](/glossary/list-constraint/) - [Boolean constraint: meaning and an example](/glossary/boolean-constraint/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/resource-lock/ --- # Organization policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud policy that sets one constraint on an organization, folder or project and applies to its descendants. It controls what resources can be configured to do, while IAM controls who can act. Rules in the policy can add conditions. ## An example, sourced Google documents that "an organization policy contains one or more rules that specify how, and whether, to enforce the constraint." One rule could enforce the constraint only on resources tagged environment=development, and another could prevent it from being enforced elsewhere. Descendants of the resource where the policy is attached inherit it. A policy is defined in a YAML or JSON file by the constraint it enforces and optionally by the conditions under which it is enforced, and each policy enforces exactly one constraint. ## Why it matters here Organization policy is the main preventive control on Google Cloud for resource configuration. Google documents both a dry-run mode and a simulator for it. Organization policy and IAM answer different questions. Use it for resource configuration rules such as locations and public IPs, and use IAM deny for permission rules. ## Related pages - [List constraint: meaning and an example](/glossary/list-constraint/) - [Boolean constraint: meaning and an example](/glossary/boolean-constraint/) - [Custom constraint: meaning and an example](/glossary/custom-constraint/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/organization-policy/ --- # List constraint Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud constraint whose policy holds a list of allowed or denied values, such as a list of permitted domains or Regions. In current Google documentation these are legacy managed constraints with list rules. Merged across levels, values from child policies combine with the parent's. ## An example, sourced Google documents that legacy managed constraints can have list rules, "a collection of allowed or denied values," and that these legacy constraints "were previously known as list constraints." The values are expressed as hierarchy subtree strings. The page names the constraint constraints/compute.trustedImageProjects as an example of one that takes a list. Google documents that when policies merge and a value conflicts, the DENY value takes precedence, and that default behavior is never merged but replaced when a policy is set. ## Why it matters here List constraints merge across the hierarchy, and deny values win in a conflict. Read the effective policy at the project before trusting an allow list set higher up. Because values merge, an allow list set high up can grow lower down. Read the effective policy on the project before trusting it. ## Related pages - [Boolean constraint: meaning and an example](/glossary/boolean-constraint/) - [Custom constraint: meaning and an example](/glossary/custom-constraint/) - [Dry-run mode: meaning and an example](/glossary/dry-run-mode/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/list-constraint/ --- # Boolean constraint Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud constraint that is either enforced or not enforced, with no list of values. A policy that sets a boolean constraint turns one behavior on or off for a resource and its descendants. It is either on or off. ## An example, sourced Google documents that legacy managed constraints can have boolean rules, which "can allow all values, deny all values, or determine if a constraint is enforced or not enforced." Custom constraints are described as functionally similar to managed constraints with boolean parameters, and are either enforced or not enforced. Managed constraints can have list or boolean parameters set by the enforcing service. ## Why it matters here Boolean constraints are the simplest to dry-run and to reason about. Several library entries, such as disabling service account key creation, are of this kind. Boolean constraints are the easiest to dry-run and reason about, since a project is either inside the rule or outside it. Several library entries, such as the one disabling service account key creation, are of this kind. ## Related pages - [Custom constraint: meaning and an example](/glossary/custom-constraint/) - [Dry-run mode: meaning and an example](/glossary/dry-run-mode/) - [IAM deny policy: meaning and an example](/glossary/iam-deny-policy/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/boolean-constraint/ --- # Custom constraint Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud constraint written by your organization, in a YAML file that names the resources, methods, conditions and actions it applies to. It covers cases that Google-managed constraints do not, within the services that support it. It is created and maintained by you. ## An example, sourced Google documents that with custom organization policies you can customize control over the specific fields your organization restricts. "A custom constraint is defined in a YAML file, which specifies the resources, methods, conditions, and actions that are subject to the constraint." Behavior and supported parameters are specific to each service, and a constraint ID must start with custom. Google lists the services that support custom constraints on a separate page. ## Why it matters here Custom constraints let you express a rule the managed set lacks. They are supported in dry-run mode and in Policy Simulator, so they can be tested before they bind. A custom constraint lets you express a rule the managed set lacks. It can be tested before enforcement, so build the test into the change process. ## Related pages - [Dry-run mode: meaning and an example](/glossary/dry-run-mode/) - [IAM deny policy: meaning and an example](/glossary/iam-deny-policy/) - [VPC Service Controls perimeter: meaning and an example](/glossary/vpc-service-controls-perimeter/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/custom-constraint/ --- # Dry-run mode Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud mode in which an organization policy is created and enforced like any other, but violations are written to audit logs and the violating actions are not denied. It shows the impact of a policy before enforcement. It is available for selected constraint types. ## An example, sourced Google documents that "an organization policy in dry-run mode is created and enforced similarly to other organization policies, and violations of the policy are audit logged, but the violating actions aren't denied." You need billing enabled for the project. Only custom constraints, managed constraints and certain legacy constraints, such as Restrict service usage, can be used in dry-run. Roles such as Organization policy administrator and Logs Configuration Writer are needed. ## Why it matters here Dry-run is the report-only phase for Google organization policy. Read the dry-run logs through a full working cycle before you set the live policy. Dry-run is the report-only phase for Google organization policy. Read its logs through a full working cycle before you set the live policy, and set a date to decide. ## Related pages - [IAM deny policy: meaning and an example](/glossary/iam-deny-policy/) - [VPC Service Controls perimeter: meaning and an example](/glossary/vpc-service-controls-perimeter/) - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/dry-run-mode/ --- # IAM deny policy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud IAM policy of deny rules attached to a project, folder or organization. Each rule names principals and permissions the principals cannot use. IAM evaluates deny policies before allow policies, so a role grant cannot override one. It is evaluated first. ## An example, sourced Google documents that deny policies "let you set guardrails on access to Google Cloud resources by creating deny rules that prevent certain principals from using certain permissions." When a principal is denied a permission, they cannot use it regardless of the roles they hold, because IAM always checks relevant deny policies before allow policies. A deny on an organization applies to every resource inside it. ## Why it matters here IAM deny is the Google Cloud equivalent of an SCP deny for permissions. A deny at the organization level cannot be loosened by a more permissive policy below it. IAM deny is Google's counterpart of an SCP deny for permissions. A deny at the organization level cannot be loosened by a more permissive policy below it, so keep exceptions in conditions. ## Related pages - [VPC Service Controls perimeter: meaning and an example](/glossary/vpc-service-controls-perimeter/) - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [Guardrail: meaning and an example](/glossary/guardrail/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/iam-deny-policy/ --- # VPC Service Controls perimeter Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud boundary around projects and supported services that controls data egress and ingress across it, independent of IAM. Requests that cross the perimeter in violation are denied, or only logged in dry run mode. It guards against data leaving a protected set of services. ## An example, sourced Google documents that with VPC Service Controls it can be difficult to determine the impact on your environment when a service perimeter is created or modified, and describes dry run mode as the answer. In dry run mode, "requests that violate the perimeter policy are not denied, only logged," and the logs can be analyzed to change the configuration. Google also says it recommends using VPC Service Controls and IAM together for defense in depth. ## Why it matters here A perimeter is the data-exfiltration guardrail on Google Cloud, separate from IAM. Run it in dry run first, because a new perimeter can block legitimate cross-project access. A perimeter protects against data leaving the boundary, which IAM alone does not. Run dry run first, since a new perimeter can block legitimate cross-project access. ## Related pages - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [Guardrail: meaning and an example](/glossary/guardrail/) - [Preventive control: meaning and an example](/glossary/preventive-control/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/dry-run-mode (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/vpc-service-controls-perimeter/ --- # Principal access boundary Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. A Google Cloud IAM policy attached to a set of principals that defines which resources those principals are eligible to access. If a principal is ineligible for a resource, access is limited whatever roles the principal holds. Its scope is the principal, not the resource. ## An example, sourced Google documents that "by default, principals are eligible to access any Google Cloud resource." With a Principal Access Boundary policy you can define the resources a principal is eligible to access. The page gives an example use: preventing principals from accessing resources in other organizations, which can help prevent phishing. These policies attach to principal sets, unlike allow policies, which attach to resources. ## Why it matters here This is Google's nearest analog of a permissions boundary, set around identities. The library covers it as a mechanism with its own entry. This is Google's nearest analog of a permissions boundary, and it is set around identities and not resources. The library covers it as its own mechanism with one entry, which you can use as a starting point. ## Related pages - [Guardrail: meaning and an example](/glossary/guardrail/) - [Preventive control: meaning and an example](/glossary/preventive-control/) - [Detective control: meaning and an example](/glossary/detective-control/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/principal-access-boundary-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/glossary/principal-access-boundary/ --- # What changed in the last 90 days Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Dated changes to preventive guardrail mechanisms and to the platforms ranked here: new policy types, changed limits, retired products and updated provider documentation and policy repositories, each linked to the page it was read from and kept in the edition record. - [VPC Service Controls quotas now show up in Cloud Quotas](/news/vpc-sc-quota-enforcement-cloud-quotas-ga/): 6 Nov 2026 - [Resource Manager adds management boundaries in preview](/news/resource-manager-management-boundaries-preview/): 30 Oct 2026 - [VPC Service Controls rules can name GKE workload identity principals](/news/vpc-sc-gke-workload-identity-ingress-egress-rules/): 14 Oct 2026 - [VPC Service Controls violation analyzer adds remediation suggestions](/news/vpc-sc-remediation-suggestions-preview/): 8 Oct 2026 - [Control Tower Account Factory for Terraform adds plan-only runs](/news/control-tower-aft-plan-only-customization-runs/): 5 Oct 2026 - [IAM Identity Center adds network access controls for Identity Store](/news/identity-center-identity-store-network-access-controls/): 5 Oct 2026 - [GuardDuty can be enabled by AWS Organizations declarative policies](/news/guardduty-declarative-policies-organization-enablement/): 1 Oct 2026 - [VPC Service Controls folder membership and folder rules reach GA](/news/vpc-sc-folders-organizations-in-rules-ga/): 30 Sep 2026 - [ECS extends CPU and memory IAM condition keys to RunTask](/news/ecs-cpu-memory-condition-keys-runtask/): 10 Sep 2026 - [VPC Service Controls perimeters can be edited with deleted principals](/news/vpc-sc-deleted-principals-in-perimeters-preview/): 3 Sep 2026 - [Lambda functions now accept full IAM resource-based policies](/news/lambda-full-iam-resource-based-policies/): 25 Aug 2026 - [IAM raises the default managed policies per role from 10 to 20](/news/iam-managed-policies-per-role-default-20/): 19 Aug 2026 - [VPC Service Controls recommender flags perimeter gaps in preview](/news/vpc-sc-recommender-perimeter-optimization-preview/): 13 Aug 2026 - [S3 access denied errors now name the SCP or RCP behind the deny](/news/s3-access-denied-messages-name-scp-rcp-policy/): 13 Aug 2026 ## Related pages - [Preventive guardrail mechanisms in AWS, Azure and Google Cloud](/mechanisms/) - [Platforms by how safely they put a guardrail into production](/platforms/) - [Picks by situation](/situations/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/ --- # VPC Service Controls quotas now show up in Cloud Quotas Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud made quota enforcement and observability for VPC Service Controls and Access Context Manager generally available on 6 November 2026. Organization administrators can see usage and limits for perimeters, protected resources, access levels and scoped access policies in the console or through the Cloud Quotas API. Google Cloud lists a VPC Service Controls feature dated 6 November 2026: quota enforcement and observability for VPC Service Controls and Access Context Manager are integrated with Cloud Quotas and are generally available. The release note says organization administrators can view quota usage and system limits for perimeters, protected resources, access levels and scoped access policies directly in the Google Cloud console or through the Cloud Quotas API. For teams that run perimeters as a preventive guardrail, the practical gain is that limits can be watched before a perimeter hits a ceiling and a change fails. Google documents the feature as generally available, so it can be added to routine monitoring. The note does not describe any change to how perimeters decide whether to allow or deny a request. ## Related pages - [Resource Manager adds management boundaries in preview](/news/resource-manager-management-boundaries-preview/) - [VPC Service Controls rules can name GKE workload identity principals](/news/vpc-sc-gke-workload-identity-ingress-egress-rules/) - [VPC Service Controls violation analyzer adds remediation suggestions](/news/vpc-sc-remediation-suggestions-preview/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-quota-enforcement-cloud-quotas-ga/ --- # Resource Manager adds management boundaries in preview Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud released management boundaries and capability configurations for Resource Manager in Preview on 30 October 2026. Tag-based or hierarchy-wide administrative perimeters span multiple projects, with BoundaryConfig, Boundary and CapabilityConfig resources set through the Google Cloud CLI, and a first capability for Agent Registry. Google Cloud lists a Resource Manager feature dated 30 October 2026: management boundaries and capability configurations are available in Preview. The release note says management boundaries establish dynamic, tag-based or hierarchy-wide administrative perimeters across multiple Google Cloud projects, so resource discovery and governance can be centralized from a designated management project, instead of being handled project by project, which Google presents as the purpose of the feature. You configure BoundaryConfig, Boundary and CapabilityConfig resources with the Google Cloud CLI to partition projects by tag values and to enable management capabilities for Agent Registry. This is an administrative grouping feature in Preview, not a data access control like an organization policy or a service perimeter, so it should be read as a governance tool and not a replacement for either. ## Related pages - [VPC Service Controls rules can name GKE workload identity principals](/news/vpc-sc-gke-workload-identity-ingress-egress-rules/) - [VPC Service Controls violation analyzer adds remediation suggestions](/news/vpc-sc-remediation-suggestions-preview/) - [Control Tower Account Factory for Terraform adds plan-only runs](/news/control-tower-aft-plan-only-customization-runs/) - [All of them](/news/) ## Sources - https://cloud.google.com/resource-manager/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/resource-manager-management-boundaries-preview/ --- # VPC Service Controls rules can name GKE workload identity principals Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud added a preview on 14 October 2026 that lets VPC Service Controls ingress and egress rules use Workload Identity Federation for GKE principals. Rules can match all pods using a Kubernetes service account, all pods in a namespace, or all pods in a cluster. Google Cloud published a VPC Service Controls entry dated 14 October 2026, marked Preview. It supports Workload Identity Federation for GKE in ingress and egress rules that allow access to resources protected by a service perimeter. The release note says the rules can name all GKE pods that use a Kubernetes service account, by name or ID, all pods in a Kubernetes namespace, or all pods in a cluster. For anyone writing perimeter exceptions, this gives a way to describe a workload by where it runs in Kubernetes instead of only by a service account or an IP range. Because the feature is in Preview, Google has not marked it for production commitments yet. Teams should test it in dry run mode before relying on it. ## Related pages - [VPC Service Controls violation analyzer adds remediation suggestions](/news/vpc-sc-remediation-suggestions-preview/) - [Control Tower Account Factory for Terraform adds plan-only runs](/news/control-tower-aft-plan-only-customization-runs/) - [IAM Identity Center adds network access controls for Identity Store](/news/identity-center-identity-store-network-access-controls/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-gke-workload-identity-ingress-egress-rules/ --- # VPC Service Controls violation analyzer adds remediation suggestions Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud added a preview on 8 October 2026 of automated remediation suggestions in the VPC Service Controls violation analyzer. It analyzes access denial events and proposes narrowly scoped ingress rules, egress rules, accessible services updates and context-aware access levels that follow least privilege. Google Cloud logged a VPC Service Controls feature in Preview on 8 October 2026: support for automated remediation suggestions in the VPC Service Controls violation analyzer. According to the release note, remediation suggestions analyze access denial events and generate actionable, narrowly scoped configuration changes to resolve perimeter denials and adhere to the principle of least privilege, as Google describes it in the note. The remediation engine can propose ingress rules, egress rules, VPC accessible services updates, and existing or new context-aware access levels. Because the output is a proposal, an administrator still decides whether to apply it. For teams that tune perimeters after a blocked request, it replaces part of the manual reading of audit logs, though the feature remains in Preview. ## Related pages - [Control Tower Account Factory for Terraform adds plan-only runs](/news/control-tower-aft-plan-only-customization-runs/) - [IAM Identity Center adds network access controls for Identity Store](/news/identity-center-identity-store-network-access-controls/) - [GuardDuty can be enabled by AWS Organizations declarative policies](/news/guardduty-declarative-policies-organization-enablement/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-remediation-suggestions-preview/ --- # Control Tower Account Factory for Terraform adds plan-only runs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 5 October 2026 that AWS Control Tower Account Factory for Terraform supports plan-only customization runs. Administrators can preview Terraform changes for global and account customizations before applying them across managed accounts, which helps with validation, drift detection and CI/CD review. AWS Control Tower Account Factory for Terraform now supports plan-only customization runs, per an AWS announcement posted 5 October 2026. AFT is an open-source Terraform module that automates account provisioning and customization in AWS Control Tower environments, and the release is tied to AFT version 1.22.0. Previously, invoking customizations always triggered a full Terraform apply, with no way to preview the changes or catch unintended drift first. With plan-only runs, administrators can trigger a Terraform plan for global and account customizations without applying anything. AWS says this supports pre-rollout validation, drift detection and CI/CD review workflows, and works with open source Terraform, Terraform Cloud and Terraform Enterprise. For teams that deliver guardrails through account customizations, the preview step lowers the risk of an unreviewed change reaching many accounts. ## Related pages - [IAM Identity Center adds network access controls for Identity Store](/news/identity-center-identity-store-network-access-controls/) - [GuardDuty can be enabled by AWS Organizations declarative policies](/news/guardduty-declarative-policies-organization-enablement/) - [VPC Service Controls folder membership and folder rules reach GA](/news/vpc-sc-folders-organizations-in-rules-ga/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/09/aws-control-tower-aft/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/control-tower-aft-plan-only-customization-runs/ --- # IAM Identity Center adds network access controls for Identity Store Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 5 October 2026 that IAM Identity Center supports network access controls for Identity Store. Administrators can require Identity Store API requests to arrive through allowed VPC endpoints or source VPCs, and limit both that API and the SCIM API to specific IP ranges. AWS IAM Identity Center now supports network access controls for Identity Store, which stores users and groups, per an AWS announcement posted 5 October 2026. You can restrict access to the Identity Store API and the SCIM API based on the network that requests come from. Custom applications use the first API, and an external identity provider uses SCIM to synchronize users and groups. For the Identity Store API you can require requests through allowed VPC endpoints or from specific source VPCs. For both APIs you can allow only specific IP ranges, with different rules for each. AWS says the controls are optional, off by default, and exempt requests that AWS services make on your behalf. They are set through the Identity Store API, SDKs and CLI. ## Related pages - [GuardDuty can be enabled by AWS Organizations declarative policies](/news/guardduty-declarative-policies-organization-enablement/) - [VPC Service Controls folder membership and folder rules reach GA](/news/vpc-sc-folders-organizations-in-rules-ga/) - [ECS extends CPU and memory IAM condition keys to RunTask](/news/ecs-cpu-memory-condition-keys-runtask/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/10/aws-identity-store-network-controls/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/identity-center-identity-store-network-access-controls/ --- # GuardDuty can be enabled by AWS Organizations declarative policies Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 1 October 2026 that Amazon GuardDuty supports AWS Organizations declarative policies. A delegated administrator can set an enablement baseline at the organization root, an OU or an account, with per-Region overrides, and enablement set by a policy cannot be changed from the GuardDuty console or API. AWS announced that Amazon GuardDuty now supports AWS Organizations declarative policies, so GuardDuty threat detection can be enabled centrally across every account and Region in an organization. The announcement, posted 1 October 2026, says the configuration applies to existing accounts and is maintained as new accounts join, with a default configuration for every Region where GuardDuty is available and optional per-Region overrides for exceptions. AWS also states that enablement set by a policy cannot be overridden through the GuardDuty console or API. For guardrail owners this extends the declarative policy model, already used for settings such as image sharing, to a security service, so account owners cannot quietly turn detection off. The delegated administrator needs permission to manage GuardDuty policies before creating one. ## Related pages - [VPC Service Controls folder membership and folder rules reach GA](/news/vpc-sc-folders-organizations-in-rules-ga/) - [ECS extends CPU and memory IAM condition keys to RunTask](/news/ecs-cpu-memory-condition-keys-runtask/) - [VPC Service Controls perimeters can be edited with deleted principals](/news/vpc-sc-deleted-principals-in-perimeters-preview/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/10/guardduty-org-enablement-policies/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/guardduty-declarative-policies-organization-enablement/ --- # VPC Service Controls folder membership and folder rules reach GA Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud announced on 30 September 2026 that VPC Service Controls now supports Google Cloud folders and organizations as resources in ingress and egress rules, and folder membership in service perimeters, both generally available. A folder member protects all projects and subfolders beneath it. Google Cloud logged two VPC Service Controls features as generally available on 30 September 2026. The first supports Google Cloud folders and organizations as resources in ingress and egress rules, which the release note says lets you create rules that allow access to and from protected resources and bypass common resource size limitations. The second supports folders as members of service perimeters. Google documents that a folder configured as a perimeter member automatically protects all projects and subfolders in that hierarchy. For a guardrail owner, that means new projects created under the folder fall inside the perimeter without a separate update. The change applies to perimeter scope only and does not change what individual IAM roles allow. ## Related pages - [ECS extends CPU and memory IAM condition keys to RunTask](/news/ecs-cpu-memory-condition-keys-runtask/) - [VPC Service Controls perimeters can be edited with deleted principals](/news/vpc-sc-deleted-principals-in-perimeters-preview/) - [Lambda functions now accept full IAM resource-based policies](/news/lambda-full-iam-resource-based-policies/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-folders-organizations-in-rules-ga/ --- # ECS extends CPU and memory IAM condition keys to RunTask Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 10 September 2026 that the ecs:task-cpu and ecs:task-memory IAM condition keys now apply to the RunTask and StartTask APIs in Amazon ECS. Administrators can enforce one consistent set of CPU and memory limits however a task is launched. Amazon ECS now supports the IAM condition keys for CPU and memory on the RunTask and StartTask APIs, per an AWS announcement dated 10 September 2026. Administrators can use them to enforce consistent CPU and memory limits across all ways of launching ECS tasks. AWS says this helps organizations prevent unexpected cost overruns and keep workloads within their resource policies. Before this change, ecs:task-cpu and ecs:task-memory worked only on RegisterTaskDefinition, CreateService and UpdateService. IAM policies that reference the keys are now evaluated when tasks start through RunTask and StartTask too. The announcement says it is available in all Regions where ECS is available at no additional cost, so an existing deny on large tasks can now cover those launch paths. ## Related pages - [VPC Service Controls perimeters can be edited with deleted principals](/news/vpc-sc-deleted-principals-in-perimeters-preview/) - [Lambda functions now accept full IAM resource-based policies](/news/lambda-full-iam-resource-based-policies/) - [IAM raises the default managed policies per role from 10 to 20](/news/iam-managed-policies-per-role-default-20/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/09/amazon-ecs-expands-condition-key-support/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/ecs-cpu-memory-condition-keys-runtask/ --- # VPC Service Controls perimeters can be edited with deleted principals Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud added a preview on 3 September 2026 so VPC Service Controls can retrieve and update service perimeters that still list deleted IAM principals. Administrators can manage those perimeters without hitting the error that the email address is invalid or non-existent. A VPC Service Controls entry dated 3 September 2026 describes a feature in Preview. VPC Service Controls supports retrieving and updating service perimeters that contain deleted IAM principals. Google says that when the feature is enabled, you can manage perimeters that contain deleted user, group or service account identities without triggering the error that the email address is invalid or non-existent. This removes a common cleanup obstacle: a perimeter rule that names a service account that no longer exists could previously block edits to the whole perimeter. The feature is in Preview, so Google may change it before general availability. Teams that keep ingress and egress rules in infrastructure as code should check how their tooling handles stale identities. ## Related pages - [Lambda functions now accept full IAM resource-based policies](/news/lambda-full-iam-resource-based-policies/) - [IAM raises the default managed policies per role from 10 to 20](/news/iam-managed-policies-per-role-default-20/) - [VPC Service Controls recommender flags perimeter gaps in preview](/news/vpc-sc-recommender-perimeter-optimization-preview/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-deleted-principals-in-perimeters-preview/ --- # Lambda functions now accept full IAM resource-based policies Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 25 August 2026 that Lambda functions support full IAM resource-based policies. One policy document can cover multiple principals and actions and use the full range of IAM condition keys, such as source IP or principal tag, instead of adding permissions one principal at a time. AWS Lambda functions now support full IAM resource-based policies, according to an announcement dated 25 August 2026. AWS says platform admins and security teams can define permissions for multiple principals and actions in a single policy document and use the full range of IAM condition keys. Previously, Lambda required permissions to be added individually per principal, which limited flexibility at scale. AWS gives examples of restricting access by source IP or principal tag and allowing several services to invoke a function with one policy. Policies can be updated in one step in the console, CLI, SDKs, CloudFormation or SAM, in all commercial Regions at no extra charge. For resource-level guardrails, conditions can now be written on Lambda the same way as on other services. ## Related pages - [IAM raises the default managed policies per role from 10 to 20](/news/iam-managed-policies-per-role-default-20/) - [VPC Service Controls recommender flags perimeter gaps in preview](/news/vpc-sc-recommender-perimeter-optimization-preview/) - [S3 access denied errors now name the SCP or RCP behind the deny](/news/s3-access-denied-messages-name-scp-rcp-policy/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/08/aws-lambda-full-iam-resource-based-policies/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/lambda-full-iam-resource-based-policies/ --- # IAM raises the default managed policies per role from 10 to 20 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 19 August 2026 that the default IAM quota for managed policies per role rose from 10 to 20. It applies automatically in all commercial Regions, GovCloud and China, and a Service Quotas request can raise the limit up to 25. AWS Identity and Access Management has increased the default quota for managed policies per role from 10 to 20, according to an announcement dated 19 August 2026. AWS says the higher default reduces the need for Service Quota requests when you follow practices such as separating permissions into purpose-specific policies or when onboarding AWS Partner products that require extra managed policies. If you need more than 20, you can request an increase up to 25 using Service Quotas. The change is available in all commercial AWS Regions, AWS GovCloud (US) and China Regions, and applies automatically to all IAM roles in an account with no action required. Teams that split permissions into smaller policies can now do so with less quota friction. ## Related pages - [VPC Service Controls recommender flags perimeter gaps in preview](/news/vpc-sc-recommender-perimeter-optimization-preview/) - [S3 access denied errors now name the SCP or RCP behind the deny](/news/s3-access-denied-messages-name-scp-rcp-policy/) - [VPC Service Controls quotas now show up in Cloud Quotas](/news/vpc-sc-quota-enforcement-cloud-quotas-ga/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/08/aws-iam-quota-increase/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/iam-managed-policies-per-role-default-20/ --- # VPC Service Controls recommender flags perimeter gaps in preview Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud introduced a preview on 13 August 2026 of a VPC Service Controls recommender that finds perimeter risks. It reports sensitive services such as BigQuery and Cloud Storage running outside perimeters, and perimeters that leave VPC accessible services unconfigured or misconfigured. Google Cloud published a VPC Service Controls entry dated 13 August 2026, marked Preview, that supports optimizing service perimeters with the VPC Service Controls recommender. The release note says it detects critical resources at risk of exfiltration, such as active and sensitive services like BigQuery and Cloud Storage operating outside service perimeters, along with unconfigured and misconfigured VPC accessible services. The recommender gives perimeter owners a list of coverage gaps instead of requiring a manual review of every project. Since the feature is in Preview, the findings are a starting point for review, not a finished audit. Google points to its page on optimizing perimeters with the recommender for details on how findings are produced. ## Related pages - [S3 access denied errors now name the SCP or RCP behind the deny](/news/s3-access-denied-messages-name-scp-rcp-policy/) - [VPC Service Controls quotas now show up in Cloud Quotas](/news/vpc-sc-quota-enforcement-cloud-quotas-ga/) - [Resource Manager adds management boundaries in preview](/news/resource-manager-management-boundaries-preview/) - [All of them](/news/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/release-notes (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/vpc-sc-recommender-perimeter-optimization-preview/ --- # S3 access denied errors now name the SCP or RCP behind the deny Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. AWS said on 13 August 2026 that Amazon S3 HTTP 403 errors now include the ARN of the specific policy that caused an explicit deny, for same-account and same-organization requests. The covered policy types are SCPs, RCPs, identity-based policies, session policies and permissions boundaries. Amazon S3 now includes the specific IAM and AWS Organizations policy Amazon Resource Name in HTTP 403 Access Denied error messages for same-account and same-organization requests, according to an AWS announcement posted 13 August 2026. Previously the message gave the policy type and reason, but with several policies of one type you still had to inspect each one by hand. For explicit deny cases the message now covers Service Control Policies, Resource Control Policies, identity-based policies, session policies and permission boundaries. AWS says the capability is available in all Regions, including GovCloud and China. For teams running preventive guardrails, this shortens the step from a blocked request to the guardrail that blocked it, which helps when a new SCP or RCP causes unexpected denials. ## Related pages - [VPC Service Controls quotas now show up in Cloud Quotas](/news/vpc-sc-quota-enforcement-cloud-quotas-ga/) - [Resource Manager adds management boundaries in preview](/news/resource-manager-management-boundaries-preview/) - [VPC Service Controls rules can name GKE workload identity principals](/news/vpc-sc-gke-workload-identity-ingress-egress-rules/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/08/s3-additional-policy-details-access-denied-error-messages/ (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/news/s3-access-denied-messages-name-scp-rcp-policy/ --- # 30 questions, answered from the data Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Thirty answers for the engineer who writes a guardrail and the lead who approves it: what an SCP can and cannot do, how SCPs, RCPs and permissions boundaries differ, Azure Policy deny against audit, Google Cloud dry run, how to test before applying, and how the platforms are ranked. **What can a service control policy do?** AWS documents an SCP as an AWS Organizations policy type that sets the maximum available permissions for IAM users and roles in member accounts. It grants nothing; the effective permissions are the intersection of the SCP with identity-based and resource-based policies. It applies to users, roles and the root user in member accounts, and AWS says SCPs affect all users and roles in attached accounts, including the root user. **What can a service control policy not do?** An SCP does not grant permissions, so an allow in an SCP does nothing on its own. AWS says SCPs do not affect users or roles in the management account, and do not affect service-linked roles. An SCP also cannot restrict a resource's access by callers outside your organization; that is the job of a resource control policy. The size limit is 10,240 characters. **What is the difference between an SCP and an RCP?** An SCP sets the maximum permissions for the IAM users and roles in member accounts. An RCP sets the maximum permissions for resources in member accounts, and applies to requests against them whether or not the caller belongs to your organization. Both grant nothing, and neither affects the management account or service-linked roles. An RCP is capped at 5,120 characters and five per root, OU or account. **What is the difference between an SCP and a permissions boundary?** An SCP is attached in AWS Organizations and caps every user and role in the accounts beneath it. A permissions boundary is a managed policy attached to a single IAM user or role, and caps what that entity's identity-based policies can grant. A boundary applies to nothing else, and works inside the account, so the two can be used together. **Does an SCP affect the management account?** No. AWS states that SCPs do not affect users or roles in the management account, and the same holds for RCPs and resources in it. A guardrail written as an SCP therefore does not protect the management account, which is one reason AWS Control Tower's documentation notes that the root user and administrators there can do work that controls would otherwise deny. **How many SCPs can I attach, and how large can they be?** AWS documents a maximum SCP size of 10,240 characters, and at most 10 SCPs attached directly to a root, OU or account, with at least one always attached. RCPs are capped at 5,120 characters and 5 per entity, counting the default RCPFullAWSAccess policy. Inside an AWS Control Tower landing zone the limit of 10 SCPs per OU also applies, so combine statements deliberately. **What is an AWS declarative policy?** A declarative policy declares and enforces a baseline configuration for an AWS service in the service's control plane, rather than regulating API access. It also covers actions by service-linked roles, which SCPs do not. For an EC2 declarative policy, account administrators cannot change the attribute and a noncompliant action fails. AWS says detaching the policy rolls the attribute back. The maximum size is 10,000 characters. **Can I test an SCP before attaching it?** Partly. AWS documents that the IAM policy simulator evaluates SCPs, including condition keys and resource scoping in deny statements, for identity-based policies, but it does not surface the matched SCP statements. AWS recommends testing an SCP on a small set of accounts in a separate OU before attaching it near the organization root. Every SCP entry in the library repeats this test step. **Can I test a resource control policy before attaching it?** Not in the IAM policy simulator, which AWS documents as not supporting RCPs. AWS suggests reviewing CloudTrail logs for Access Denied errors before attaching one. Because an RCP applies to callers outside your organization too, the review should include partners and third-party integrations. IAM Access Analyzer reads RCPs in its analysis, and picks up a change at its next periodic scan within 24 hours. **What is the difference between Azure Policy deny and audit?** Deny stops a matching create or update request before it reaches the resource provider. Audit lets the request through and creates a warning event. Microsoft documents that deny does not touch existing resources, which are only marked noncompliant. Modify and deployIfNotExists are different again: they change or deploy settings, and existing resources change only through a remediation task. Assigning with audit first is the lower-risk start. **How do I test an Azure Policy before enforcing it?** Assign the definition with enforcementMode set to DoNotEnforce, which Microsoft documents as the way to evaluate an assignment before enforcement, or assign it with the audit effect first. Review the compliance results, fix or exempt what shows up, then switch the mode. Remember that effects are evaluated in order, so an earlier modify or append can change the request before a deny is considered. **Can I create Azure deny assignments myself?** Microsoft documents that customer-managed deny assignments are in private preview, and that for generally available scenarios Azure creates and manages them, for example when you set deny settings on a deployment stack. A stack can exclude at most five principals and its deny settings cover only resources the template defines explicitly. No dry run was found for deny assignments on 2026-11-10. **Should I use an Azure resource lock or a policy?** They solve different problems. A lock blocks deletion (CanNotDelete) or modification and deletion (ReadOnly) of a scope, and overrides user permissions, but only for control plane operations. A ReadOnly lock can block a virtual machine from starting, because that uses POST. A policy evaluates the content of a request. Locks cannot be added to management groups, and Owners can remove them. **How does Google Cloud organization policy dry run work?** Google documents organization policies in dry-run mode: violations are audit logged but not denied, so you can see what a constraint would block before enforcing it. Dry run covers custom constraints, managed constraints and a short list of legacy managed constraints. Policy Simulator for Organization Policy also lists the resources that would violate a proposed policy, but it cannot test changes to legacy managed constraints. **Are Google Cloud organization policies retroactive?** Usually not. Google says new organization policies are usually not retroactive, so existing violations are not stopped, and the library notes cases such as Cloud SQL public IP access and Vertex AI Workbench settings, which apply only to new instances. Cloud Storage public access prevention is the exception in the library: Google documents that it revokes public access for existing buckets and objects once enabled. **In what order does Google Cloud check deny and allow policies?** Google documents that IAM checks relevant deny policies before allow policies, so a deny blocks a permission regardless of the roles granted. Principal access boundary policies work the same way: they define the resources a principal is eligible to access and can only block. Policy Troubleshooter examines the allow, deny and principal access boundary policies together and says whether a principal can use a permission. **What does VPC Service Controls not block?** Google documents that a service perimeter blocks communication to Google Cloud managed services across the boundary, but not third-party APIs or services on the internet, and that not all Google Cloud services are supported. Changes can take up to 30 minutes to propagate and may briefly block requests. Dry run mode logs requests that would violate the perimeter without denying them. **What is a custom organization policy constraint?** It is a constraint you define in a YAML file that names the resources, methods, conditions and actions, with the condition written in Common Expression Language (CEL). You then enforce it with an organization policy. Most resource types support up to 20 custom constraints, a condition holds up to 1,500 characters, and a policy set with set-policy can take up to 15 minutes to take effect. **What is the difference between a preventive and a detective control?** A preventive control answers a request with a denial before it takes effect: an SCP, an Azure Policy deny, a Google Cloud constraint. A detective control reports after the fact. AWS Control Tower shows the split: preventive controls use SCPs and RCPs, detective controls use AWS Config, and proactive controls use CloudFormation hooks, which only cover CloudFormation stack operations. **How do I test a guardrail before applying it?** Use the mechanism's own test first: the IAM policy simulator and a separate test OU in AWS, enforcementMode DoNotEnforce or the audit effect in Azure, dry-run mode and Policy Simulator in Google Cloud. Then read the log for the period before you enforce. The Simulation Coverage Matrix shows which platforms add a test of their own, and each library entry names the test step and its source. **Where do the 100 guardrails come from?** From provider documentation and from open source sets with permissive licenses. Sixty are taken from the aws-samples service control policy and resource control policy repositories (MIT-0) and the Azure/azure-policy repository (MIT License), and 40 are written from provider documentation. Nothing is invented, each entry links to its file or page, and the dates read are 10 Nov 2026. **Can I copy the policies in the library?** The aws-samples repositories state MIT-0, and the LICENSE file read is MIT-0 text, copyright Amazon.com, Inc. The Azure/azure-policy repository is under the MIT License. Entries written from provider documentation are examples of what the documentation describes. Each entry states its license, and every policy needs testing in your environment first, since the placeholders in many must be replaced. **How are the platforms ranked here?** On seven criteria with weights fixed before any score existed: safe to enforce 30, one change for a class of risk 30, removes access 15, stays fixed 10, uses the cloud's own controls 10, evidence for auditors 3, and identity, network and AI coverage 2. The total is a weighted mean rounded to one decimal. Act leads at 5.9, then Sonrai Security at 5.7 and AWS Control Tower at 4.2. **Why does Act rank first?** Act scores 7 on safe to enforce, the only 7 and the criterion with the largest weight. Act says each change is tested against a digital twin built from cloud configuration data, then run in audit mode against audit-log events. It also scores 7 on one change for a class of risk, with guardrail campaigns at the SCP level. Act is scored by the same rules as every other platform. **Where is Act behind?** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. On removes, Act scores 3 against 5 for Sonrai. **Why do Sonrai and Act tie on regulated and small-team?** Both total 5.2 on the regulated view and 5.5 on the small-team view at one decimal, so they share first place. Sonrai is listed first because it is ahead before rounding: its removes lead, and its edge on native controls and audit evidence, slightly exceed Act's lead on safe to enforce and on identity, network and AI coverage. Sonrai's pages say it deploys the blocking policy and updates it on approval; Act delivers a pull request for your team to apply. **Why is AWS Control Tower third?** Control Tower scores 4.2. It applies preventive controls to whole OUs from a catalog of more than 750 controls through SCPs and RCPs, and scans for drift daily. It is AWS only, has a management-account bypass, is limited to 10 SCPs per OU, and AWS tells users to test proactive controls themselves, with no impact simulation described. That is why it scores 1 on safe to enforce. **Does any platform simulate a guardrail against my real environment?** Act and Google Cloud's Policy Simulator are the clearest cases in the evidence. Act says it tests changes on a digital twin of your configuration and in audit mode on your audit logs. Google lists resources that would violate a proposed organization policy before enforcement, for Google Cloud only. CrowdStrike states a dry run without saying what it runs against. Sonrai shows its exemption list before deployment but is not described as simulating. **Do the just-in-time access tools write guardrails?** Not in the evidence read. Apono can place a dormant principal in an AWS permission boundary and gives a deny rule for the user to create. Britive scopes access by profile; a deny or boundary: none found on 2026-11-10. P0 has a deny-all policy that stops requests through P0 and not a cloud boundary. Entitle sets approval workflow, owner and requestable status. They are ranked here on that basis. **What happened to Microsoft's CIEM product?** Microsoft is scored on cloud infrastructure entitlement management in Defender for Cloud, because Entra Permissions Management was retired on 2025-11-01 and has not been sold since 2025-04-01. Defender for Cloud gives recommendations described as guidance for remediation. Azure Policy is a separate Azure service, and Microsoft's pages do not describe it acting on CIEM findings. ## Related pages - [The Guardrail Library: the data, downloadable](/datasets/) - [The Guardrail Library: 100 preventive guardrails for AWS, Azure and Google Cloud](/library/) - [The Simulation Coverage Matrix: who shows a guardrail's effect before it ships](/simulation/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/faq/ --- # The Guardrail Library: 100 preventive guardrails as a dataset, with the policy, what it blocks and what it breaks Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The dataset holds 100 entries: 40 for AWS, 30 for Azure and 30 for Google Cloud. Each has its mechanism, risk class, policy text, what it blocks, what it breaks with a quote, how to test it before applying, and the source file with its license. Edition 10 Nov 2026. Each record has a stable identifier, a title, the mechanism (for example scp, azure-policy or org-policy), one of ten risk classes, the policy in its own language, and the license of the source. Sixty records come from MIT-0 and MIT licensed repositories, and 40 from provider documentation. Every record also carries the read date, 10 Nov 2026. The dataset is for engineers who want to start from a tested example and for leads who want to see what a guardrail costs before approving it. The "breaks" field is the reason it exists: it records, with a quote, what the source says the policy stops that someone may have needed. ## Downloads - [The Guardrail Library (CSV)](/datasets/guardrail-library-2026-11-10.csv): 100 rows, edition 10 Nov 2026 - [The Guardrail Library (JSON)](/datasets/guardrail-library-2026-11-10.json): 100 preventive guardrails for AWS, Azure and Google Cloud: policy text, what each blocks and breaks, the test method, the source and license. - [Simulation Coverage Matrix (CSV)](/datasets/simulation-coverage-2026-11-10.csv): 29 rows, edition 10 Nov 2026 - [Simulation Coverage Matrix (JSON)](/datasets/simulation-coverage-2026-11-10.json): The 16 platforms with their safe to enforce score and the evidence about testing a guardrail first, plus each cloud mechanism's native test method. - [Platforms by how safely they put a guardrail into production (CSV)](/datasets/guardrail-platforms-2026-11-10.csv): 16 rows, edition 10 Nov 2026 - [Platforms by how safely they put a guardrail into production (JSON)](/datasets/guardrail-platforms-2026-11-10.json): 16 platforms scored 0 to 10 on seven criteria with this site's weights. ## Cite this **Cite this.** Cite this line: 'Cloud Governance Tools, The Guardrail Library, edition 10 Nov 2026, https://cloudgovernancetools.com/datasets/' ## Changelog - 10 Nov 2026: first edition. Every row read from the source linked in the row on 10 Nov 2026. ## Related pages - [The Guardrail Library: 100 preventive guardrails for AWS, Azure and Google Cloud](/library/) - [The Simulation Coverage Matrix: who shows a guardrail's effect before it ships](/simulation/) - [AWS guardrails: 40 preventive policies with what they block and break](/library/aws/) ## Sources - Deny member accounts leaving the organization: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-member-accounts-from-leaving-your-AWS-organization.json (read 10 Nov 2026) - Prevent root credentials management in member accounts: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-root-credentials-management-in-member-accounts-in-AWS-Organizations.json (read 10 Nov 2026) - Deny the root user except for S3 bucket policy changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-the-root-user-from-performing-actions-except-S3-bucketpolicy-changes.json (read 10 Nov 2026) - Service control policies (SCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - Resource control policies (RCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - IAM permissions boundaries: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - act.security/platform: https://act.security/platform (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/datasets/ --- # Cloud Governance Tools documents preventive guardrails from provider sources and ranks the platforms that deploy them Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Cloud Governance Tools is a reference with two parts: 100 preventive guardrails for AWS, Azure and Google Cloud, each with what it blocks and breaks, and sixteen platforms ranked on how safely they put a guardrail into production. Every fact carries a quote, a link and the date it was read. The library part is built from provider documentation and from open source policy sets published under permissive licenses. The platform part is built from each vendor's own pages and from analyst reports about the vendor. Nothing is invented: where a page was read and a capability was not found, the page says "none found on 2026-11-10". The ranking method, weights and scores are published, and the weights were fixed before any score existed. Act is a platform in the ranking and is scored by the same rules as the others. The site covers preventive guardrails only; cloud security posture rankings, workforce identity products and agent governance are other subjects and are not ranked here. ## Related pages - [Editorial policy](/editorial-policy/) - [Corrections](/corrections/) - [Contact](/contact/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/about/ --- # How Cloud Governance Tools is written Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Every statement is either a fact with a quote, a link and a read date, or a score derived from those facts by published weights. A vendor's claim about itself is written as the vendor's claim, such as "Act says". A capability not found in the pages read is written as "none found on 2026-11-10". Policies in the library are copied from their sources with the license noted, or written from provider documentation. The "breaks" field is taken from the source's own README or from the provider's documentation, with the quote. Scores are integers from 0 to 10 per criterion. The criteria, the weights and the vendor set were committed before the first score. When a score changes, the change is dated on the corrections page with the reason. ## Related pages - [Corrections](/corrections/) - [Contact](/contact/) - [Privacy](/privacy/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/editorial-policy/ --- # Corrections and the correction rule Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Corrections are dated and public. A fact that is wrong, out of date or taken from the wrong page is corrected in the data, the affected scores are recomputed under the same weights, and the change is recorded here with its date and reason. Send a correction with the page it concerns and the source that shows the fix. Read dates matter: the edition of 10 Nov 2026 reflects pages as they stood that day. Provider documentation changes, and a guardrail that was correct on that date may need a new check. A vendor may ask for a fact to be re-read; the new fact replaces the old one only with a new quote, link and date. No correction has been published since the first edition of 10 Nov 2026. Before that edition the scores went through three blind scoring rounds and three reviews; 76 score changes were logged, each with the fact that caused it, its source and its read date, and the scores printed on this site are the final ones. From here on, every change to a published score is listed on this page with the old value, the new value, the fact and the date. ## Related pages - [Contact](/contact/) - [Privacy](/privacy/) - [Terms](/terms/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/corrections/ --- # Contact Cloud Governance Tools Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Write to editor@cloudgovernancetools.com with a correction, a source we should read, or a question about a score. Include the page address, the fact or entry concerned and, for a correction, the provider or vendor page that shows the fix, and we will answer with the date the data was re-read. Messages about a platform's score are answered with the facts behind it. The scores follow the published weights, so a change needs a new fact with a quote and a link, not a request. ## Related pages - [Privacy](/privacy/) - [Terms](/terms/) - [How platforms are ranked on putting a guardrail into production](/method/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/contact/ --- # Privacy Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The site sets no cookies and runs no analytics script. The host keeps ordinary server logs of requests (address, time, page, status), used to keep the site running and to find errors, and no profile of any visitor is built from them. Email sent to the editor address is used to answer it and to record corrections. It is not added to a list and not passed on. ## Related pages - [Terms](/terms/) - [How platforms are ranked on putting a guardrail into production](/method/) - [30 questions about cloud governance guardrails](/faq/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/privacy/ --- # Terms of use Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The site is a reference, offered as is. Policies in the library are examples from their sources, under the licenses stated on each entry (MIT-0 for the aws-samples repositories, MIT for Azure/azure-policy), and must be tested in your own environment before they are applied. Scores describe documentation read on a stated date. Text on the site may be quoted with a link to the page and its edition date. The licenses of the source policies govern the policies themselves. Vendor and provider names are used to identify products and belong to their owners. ## Related pages - [How platforms are ranked on putting a guardrail into production](/method/) - [30 questions about cloud governance guardrails](/faq/) - [The Guardrail Library: the data, downloadable](/datasets/) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/terms/ --- # The Guardrail Library: 100 preventive guardrails, 40 for AWS, 30 for Azure and 30 for Google Cloud Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 100 preventive guardrails: 40 for AWS, 30 for Azure and 30 for Google Cloud. By risk class it counts 19 AI services, 14 identity and privilege, 13 encryption and keys, 10 public exposure, 9 organization integrity, 8 data exfiltration, 8 logging, 7 network, 7 resource and cost, and 5 region. Every entry comes from a source that can be checked, and nothing is invented. Sixty entries are taken from open source sets under their licenses: 33 for AWS from the aws-samples service-control-policy-examples and resource-control-policy-examples repositories (27 SCPs and 6 RCPs, MIT-0, with the LICENSE file read) and 27 for Azure from the Azure/azure-policy repository (MIT License). The other 40 are written from provider documentation: 7 for AWS, 3 for Azure and 30 for Google Cloud, where the entries are organization policy constraints and other mechanisms Google documents. Each entry gives the policy, the mechanism it uses, what it blocks, what it breaks, and how to test it before applying. "Breaks" is taken from the source's own README or from the provider's documentation, with the quote and the page it came from. Where the provider documents no way to test a guardrail, the entry says "none found on 2026-11-10". Entries were read on 10 Nov 2026, and the source link on each entry goes to the exact file or page. ## By risk class - [Identity and privilege](/risk/identity-and-privilege/): 14 guardrails - [Data exfiltration](/risk/data-exfiltration/): 8 guardrails - [Public exposure](/risk/public-exposure/): 10 guardrails - [Logging and monitoring](/risk/logging-and-monitoring/): 8 guardrails - [Encryption and keys](/risk/encryption-and-keys/): 13 guardrails - [Network perimeter](/risk/network-perimeter/): 7 guardrails - [Region and residency](/risk/region-and-residency/): 5 guardrails - [Organization integrity](/risk/organization-integrity/): 9 guardrails - [AI services](/risk/ai-services/): 19 guardrails - [Resource and cost](/risk/resource-and-cost/): 7 guardrails ## Every entry | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Network perimeter | documented | | [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny IAM and account actions invoked from chat applications](/guardrails/aws-scp-deny-chat-app-iam-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Data exfiltration | documented | | [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Require TLS 1.2 or later for S3](/guardrails/aws-rcp-s3-minimum-tls-1-2/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Deny S3 customer-provided encryption keys (SSE-C)](/guardrails/aws-rcp-deny-s3-sse-c/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Organization integrity | documented | | [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Network perimeter | documented | | [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Data exfiltration | documented | | [Require IMDSv2 for new EC2 instances (declarative policy)](/guardrails/aws-declarative-require-imdsv2/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Identity and privilege | documented | | [Enforce an Amazon Bedrock guardrail on model inference (declarative policy)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | none found | | [Opt out of AI service content use organization-wide (declarative policy)](/guardrails/aws-declarative-ai-services-opt-out/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | documented | | [Permissions boundary limiting a user to S3, CloudWatch and EC2](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | | [Permissions boundary for delegated IAM user creation](/guardrails/aws-pb-delegated-user-creation-boundary/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | | [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Region and residency | documented | | [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Organization integrity | documented | | [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Data exfiltration | documented | | [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure Machine Learning registry deployments restricted to the allowed registry (preview)](/guardrails/azure-policy-ml-registry-models/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Microsoft Foundry agents and toolboxes should not use denied tools (preview)](/guardrails/azure-policy-foundry-denied-tools/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure AI Search services should disable public network access](/guardrails/azure-policy-search-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/) | Azure | [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/) | Organization integrity | none found | | [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/) | Azure | [Azure resource locks](/mechanisms/resource-lock/) | Organization integrity | none found | | [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/) | Azure | [Management group assignment](/mechanisms/management-groups/) | Region and residency | documented | | [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Identity and privilege | documented | | [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/) | Google Cloud | [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/) | Data exfiltration | documented | | [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/) | Google Cloud | [Custom organization policy constraints](/mechanisms/custom-org-policy/) | Data exfiltration | documented | | [Principal access boundary limiting the resources principals can access](/guardrails/gcp-pab-eligible-resources/) | Google Cloud | [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/) | Data exfiltration | documented | | [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | | [Restrict VPC peering usage](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | | [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | | [Disable Audit Logging exemption](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | | [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Region and residency | none found | | [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | documented | | [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | none found | | [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Organization integrity | documented | | [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Restrict egress domains for data connectors](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | documented | | [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | | [Restrict load balancer creation based on load balancer types](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## Cite this **Cite this.** Cite this line: 'Cloud Governance Tools, The Guardrail Library, edition 10 Nov 2026, https://cloudgovernancetools.com/datasets/' **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [The Simulation Coverage Matrix: who shows a guardrail's effect before it ships](/simulation/) - [AWS guardrails: 40 preventive policies with what they block and break](/library/aws/) - [Azure guardrails: 30 preventive policies with what they block and break](/library/azure/) ## Sources - AWS: Require IAM authentication on Lambda function URLs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-modification-to-Lambda-URL-Config.json (read 10 Nov 2026) - AWS: Restrict access to resources in a suspended OU: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/Deny-resource-access-if-the-resource-belongs-to-a-specific-organizational-unit.json (read 10 Nov 2026) - AWS: Opt out of AI service content use organization-wide (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out_syntax.html (read 10 Nov 2026) - AWS: Permissions boundary for delegated IAM user creation: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Azure: Azure AI Search services should disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Search/RequirePublicNetworkAccessDisabled_Deny.json (read 10 Nov 2026) - Azure: Deployment stack deny settings (Azure RBAC deny assignment): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - Azure: Resource locks (CanNotDelete and ReadOnly): https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) - Azure: Allowed locations assigned at a management group: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/library/ --- # AWS offers four preventive mechanisms, each evaluated at a different layer Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 40 AWS guardrails across 10 risk classes, written in Service control policies (SCPs), Resource control policies (RCPs), IAM permissions boundaries, Declarative policies. 39 of them have a documented way to test the effect before enforcing. Each entry copies the policy text from AWS documentation or a licensed open source set and says what it blocks and breaks. AWS documents SCPs and RCPs as AWS Organizations policy types that set the maximum available permissions, and both grant nothing. An SCP caps the IAM users, roles and root user in member accounts. An RCP caps the resources of supported services in member accounts, for any principal calling them, whether or not the caller belongs to the organization. Neither affects the management account or service-linked roles. Evaluation is by intersection: SCPs and RCPs are evaluated together with identity-based and resource-based policies, and the effective permissions are what all layers allow. A permissions boundary works inside one account, on one IAM user or role, and caps what that entity's identity-based policies can grant. Declarative policies are different in kind: they enforce a baseline configuration in the service control plane, including for service-linked roles, so a noncompliant action fails, and detaching the policy rolls the attribute back. The library holds 40 AWS entries: 27 SCPs, 6 RCPs, 5 declarative policies and 2 permissions boundaries. ## The mechanisms - [Service control policies (SCPs)](/mechanisms/scp/): AWS documents SCPs as an AWS Organizations policy type that sets the maximum available permissions for IAM users and roles in member accounts. - [Resource control policies (RCPs)](/mechanisms/rcp/): AWS documents RCPs as an AWS Organizations policy type that sets the maximum available permissions for resources in member accounts. - [IAM permissions boundaries](/mechanisms/permissions-boundary/): AWS documents a permissions boundary as a managed policy attached to an IAM user or role that sets the maximum permissions that an identity-based policy can grant to that entity. - [Declarative policies](/mechanisms/declarative-policy/): AWS documents declarative policies as an AWS Organizations policy type that declares and enforces a baseline configuration for an AWS service in the service's control plane, rather than regulating API access the way SCPs and RCPs do. ## The 40 entries | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Network perimeter | documented | | [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny IAM and account actions invoked from chat applications](/guardrails/aws-scp-deny-chat-app-iam-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Data exfiltration | documented | | [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Require TLS 1.2 or later for S3](/guardrails/aws-rcp-s3-minimum-tls-1-2/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Deny S3 customer-provided encryption keys (SSE-C)](/guardrails/aws-rcp-deny-s3-sse-c/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Organization integrity | documented | | [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Network perimeter | documented | | [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Data exfiltration | documented | | [Require IMDSv2 for new EC2 instances (declarative policy)](/guardrails/aws-declarative-require-imdsv2/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Identity and privilege | documented | | [Enforce an Amazon Bedrock guardrail on model inference (declarative policy)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | none found | | [Opt out of AI service content use organization-wide (declarative policy)](/guardrails/aws-declarative-ai-services-opt-out/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | documented | | [Permissions boundary limiting a user to S3, CloudWatch and EC2](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | | [Permissions boundary for delegated IAM user creation](/guardrails/aws-pb-delegated-user-creation-boundary/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Azure guardrails: 30 preventive policies with what they block and break](/library/azure/) - [Google Cloud guardrails: 30 preventive policies with what they block and break](/library/gcp/) - [About](/about/) ## Sources - Service control policies (SCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - Resource control policies (RCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - IAM permissions boundaries: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Declarative policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/library/aws/ --- # Azure blocks at create and update, and a deny does not touch what already exists Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 30 Azure guardrails across 10 risk classes, written in Azure Policy (built-in definitions), Azure RBAC deny assignments (created by deployment stacks), Azure resource locks, Management group assignment. 28 of them have a documented way to test the effect before enforcing. Azure Policy evaluates create and update requests against assigned definitions. With the deny effect a matching request is stopped before it reaches the resource provider, while modify and deployIfNotExists change or deploy settings instead of blocking, and audit only warns. Effects are evaluated in order, so an earlier modify or append can change the request so a deny does not trigger. Existing resources that match are only marked noncompliant. Assignments cascade from management groups by inheritance, and scopes excluded with notScopes are not evaluated at all. Three other mechanisms sit beside it. Resource locks block deletion or modification at a subscription, resource group or resource and override user permissions, but they apply only to control plane operations. Deny assignments attach deny actions to principals at a scope; Azure creates them for generally available scenarios such as deployment stack deny settings, and customer-managed deny assignments are in private preview. Management groups organize subscriptions into a hierarchy up to six levels deep. The library holds 30 Azure entries, 27 of them Azure Policy. ## The mechanisms - [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Azure Policy evaluates resource create and update requests against assigned policy definitions. - [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/): A deny assignment attaches a set of deny actions to a user, group or service principal at a scope. - [Azure resource locks](/mechanisms/resource-lock/): A lock on a subscription, resource group or resource blocks deletion (CanNotDelete) or modification and deletion (ReadOnly) of the locked scope. - [Management group assignment](/mechanisms/management-groups/): Management groups organize subscriptions into a hierarchy so that policy and access assigned on a management group cascade to all its subscriptions by inheritance. ## The 30 entries | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Region and residency | documented | | [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Organization integrity | documented | | [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Data exfiltration | documented | | [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure Machine Learning registry deployments restricted to the allowed registry (preview)](/guardrails/azure-policy-ml-registry-models/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Microsoft Foundry agents and toolboxes should not use denied tools (preview)](/guardrails/azure-policy-foundry-denied-tools/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure AI Search services should disable public network access](/guardrails/azure-policy-search-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/) | Azure | [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/) | Organization integrity | none found | | [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/) | Azure | [Azure resource locks](/mechanisms/resource-lock/) | Organization integrity | none found | | [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/) | Azure | [Management group assignment](/mechanisms/management-groups/) | Region and residency | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Google Cloud guardrails: 30 preventive policies with what they block and break](/library/gcp/) - [About](/about/) - [Editorial policy](/editorial-policy/) ## Sources - Azure Policy (built-in definitions): https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - Azure RBAC deny assignments (created by deployment stacks): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - Azure resource locks: https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) - Management group assignment: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/library/azure/ --- # Google Cloud layers five mechanisms, with IAM deny checked before allow and dry run on most Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The library holds 30 Google Cloud guardrails across 10 risk classes, written in Organization Policy Service (managed and legacy constraints), Custom organization policy constraints, IAM deny policies, VPC Service Controls (service perimeters), Principal access boundary (PAB) policies. 7 of them have a documented way to test the effect before enforcing. Google Cloud's organization policies enforce constraints on an organization, folder or project, and descendants inherit them by default. Custom constraints are written in CEL and enforced with an organization policy. IAM deny policies are checked before allow policies, and block named principals from named permissions regardless of the roles they hold. Principal access boundary policies define the resources a set of principals is eligible to access, and can only block, never grant. VPC Service Controls perimeters block communication to managed Google Cloud services across the boundary by default. Testing is built into more of this stack than in the other clouds. Organization policies and VPC Service Controls both have dry-run modes that log violations without denying them, and Policy Simulator and Policy Troubleshooter cover other policy types. Limits apply: new organization policies are usually not retroactive, dry run excludes most legacy managed constraints, custom constraint changes take up to 15 minutes, and perimeter changes up to 30. The library holds 30 Google Cloud entries, 25 of them organization policy. ## The mechanisms - [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Organization policies set restrictions on Google Cloud resources by enforcing a constraint on an organization, folder or project. - [Custom organization policy constraints](/mechanisms/custom-org-policy/): A custom constraint is defined in a YAML file that names the resources, methods, conditions and actions subject to the constraint. - [IAM deny policies](/mechanisms/iam-deny/): Deny policies are made up of deny rules that prevent certain principals from using certain permissions, regardless of the roles they are granted. - [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/): A service perimeter creates a security boundary around Google Cloud resources. - [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/): A principal access boundary policy defines the resources a set of principals is eligible to access. ## The 30 entries | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Identity and privilege | documented | | [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/) | Google Cloud | [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/) | Data exfiltration | documented | | [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/) | Google Cloud | [Custom organization policy constraints](/mechanisms/custom-org-policy/) | Data exfiltration | documented | | [Principal access boundary limiting the resources principals can access](/guardrails/gcp-pab-eligible-resources/) | Google Cloud | [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/) | Data exfiltration | documented | | [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | | [Restrict VPC peering usage](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | | [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | | [Disable Audit Logging exemption](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | | [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Region and residency | none found | | [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | documented | | [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | none found | | [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Organization integrity | documented | | [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Restrict egress domains for data connectors](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | documented | | [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | | [Restrict load balancer creation based on load balancer types](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [About](/about/) - [Editorial policy](/editorial-policy/) - [Corrections](/corrections/) ## Sources - Organization Policy Service (managed and legacy constraints): https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - Custom organization policy constraints: https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints (read 10 Nov 2026) - IAM deny policies: https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) - VPC Service Controls (service perimeters): https://cloud.google.com/vpc-service-controls/docs/overview (read 10 Nov 2026) - Principal access boundary (PAB) policies: https://cloud.google.com/iam/docs/principal-access-boundary-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/library/gcp/ --- # Identity and privilege guardrails cap what administrators can do, and break the tasks that need those rights Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 14 guardrails in the library address identity and privilege: 7 AWS, 3 Azure, 4 Google Cloud. The first in the list, "Prevent root credentials management in member accounts", denies all actions for the member account root user unless the session was created with sts:AssumeRoot (the aws:AssumedRoot key is not null), so root credentials are managed centrally from the management account. Guardrails in this class restrict the root user, IAM user credentials, federation providers and role creation, and set permissions boundaries. The library holds 14, 7 for AWS. They break the work they restrict. Denying the root user except for a few tasks makes every other task done as root in an affected account fail, and the README points to the AWS list of tasks that require root credentials. Denying changes to IAM federation providers blocks teams that manage their own identity providers unless they use the privileged role. A permissions boundary can make an iam:CreateUser call fail because the boundary does not allow the action. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Identity and privilege | documented | | [Require IMDSv2 for new EC2 instances (declarative policy)](/guardrails/aws-declarative-require-imdsv2/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Identity and privilege | documented | | [Permissions boundary limiting a user to S3, CloudWatch and EC2](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | | [Permissions boundary for delegated IAM user creation](/guardrails/aws-pb-delegated-user-creation-boundary/) | AWS | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | Identity and privilege | documented | | [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Identity and privilege | documented | | [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Identity and privilege | none found | | [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Identity and privilege | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Prevent root credentials management in member accounts** (AWS): Direct root user sign-in in a member account can no longer perform actions; the repository README notes the SCP restricts the member account root user and has no effect on the management account. - **Deny the root user except for S3 bucket policy changes** (AWS): Any other task done as the root user in an affected account fails; the README points to the AWS list of tasks that require root user credentials. - **Deny critical IAM user actions** (AWS): Anyone outside the named privileged role or tag exception cannot create or rotate IAM user credentials, and the placeholder [PRIVILEGED_ROLE] must be replaced before use. - **Deny creating or changing IAM federation providers** (AWS): Teams that manage their own identity providers in member accounts are blocked unless they use the privileged role; the README says it is usually paired with the critical IAM user actions policy. - **Require IMDSv2 for new EC2 instances (declarative policy)** (AWS): AWS says to make sure no instances are making IMDSv1 calls before setting http_tokens to required, and recommends a hop limit of at least 2 when it is required. - **Permissions boundary limiting a user to S3, CloudWatch and EC2** (AWS): AWS gives the example of an iam:CreateUser permission policy failing because the boundary does not allow the action. - **Permissions boundary for delegated IAM user creation** (AWS): AWS notes the delegated administrator can create a user who can access S3 resources the administrator cannot, so delegating effectively trusts the administrator with that access. - **Storage accounts should prevent shared key access** (Azure): Existing storage accounts that allow Shared Key are marked non-compliant but not changed. - **Azure Key Vault should use the RBAC permission model** (Azure): Existing key vaults on access policies are marked non-compliant but not changed. - **Azure SQL Database should have Microsoft Entra-only authentication enabled** (Azure): The definition does not block servers from being created with local authentication enabled; Microsoft points to the Entra-only authentication initiative to cover both. - **Disable service account key creation** (Google Cloud): Google documents that user-managed credentials cannot be created for service accounts in affected projects, so tools or workloads that depend on newly created keys cannot get them. - **Disable automatic IAM grants for default service accounts** (Google Cloud): Google documents that it does not stop the Editor or Owner role being granted to default service accounts later, and that other roles must be granted instead of Editor. - **Require OS Login** (Google Cloud): Google documents that setting OS Login metadata deletes the VM authorized_keys files and VMs no longer accept SSH keys stored in project or instance metadata. - **Deny service account key create and delete to a group on one project** (Google Cloud): Google documents that a denied principal cannot use the permission whatever roles they hold, so an exception group (exceptionPrincipals) is needed for anyone who must keep the access. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Data exfiltration: 8 preventive guardrails across three clouds](/risk/data-exfiltration/) - [Public exposure: 10 preventive guardrails across three clouds](/risk/public-exposure/) - [Logging and monitoring: 8 preventive guardrails across three clouds](/risk/logging-and-monitoring/) ## Sources - Prevent root credentials management in member accounts: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-root-credentials-management-in-member-accounts-in-AWS-Organizations.json (read 10 Nov 2026) - Deny the root user except for S3 bucket policy changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-the-root-user-from-performing-actions-except-S3-bucketpolicy-changes.json (read 10 Nov 2026) - Deny critical IAM user actions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-critical-IAM-user-actions.json (read 10 Nov 2026) - Deny creating or changing IAM federation providers: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-unwarranted-IAM-federations-creation-modification.json (read 10 Nov 2026) - Require IMDSv2 for new EC2 instances (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - Permissions boundary limiting a user to S3, CloudWatch and EC2: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Permissions boundary for delegated IAM user creation: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Storage accounts should prevent shared key access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountAllowSharedKeyAccess_Audit.json (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/identity-and-privilege/ --- # Data exfiltration guardrails stop sharing and exports, and they break legitimate sharing Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 8 guardrails in the library address data exfiltration: 2 AWS, 1 Azure, 5 Google Cloud. The first in the list, "Deny AWS RAM sharing with external accounts", denies ram:CreateResourceShare and ram:UpdateResourceShare when the share allows external principals. Each entry gives the policy text, what it blocks, what it is known to break and how to test it first. Guardrails in this class stop data from leaving the organization: resource sharing with outside accounts, public snapshots, long-lived access tokens, exports to outside destinations and service perimeters. The library holds 8 of them, 5 for Google Cloud. What they break is legitimate sharing. In AWS, denying RAM sharing with external accounts makes sharing a subnet fail for every principal in the affected accounts. In Google Cloud, domain restricted sharing needs your own organization principal set and Google Workspace ID listed, or your own principals cannot be granted roles. Google also says moves within an organization are not governed by the export constraint. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Data exfiltration | documented | | [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Data exfiltration | documented | | [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Data exfiltration | documented | | [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/) | Google Cloud | [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/) | Data exfiltration | documented | | [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Data exfiltration | none found | | [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/) | Google Cloud | [Custom organization policy constraints](/mechanisms/custom-org-policy/) | Data exfiltration | documented | | [Principal access boundary limiting the resources principals can access](/guardrails/gcp-pab-eligible-resources/) | Google Cloud | [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/) | Data exfiltration | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny AWS RAM sharing with external accounts** (AWS): Sharing resources such as VPC subnets with accounts outside the organization fails for every principal in the affected accounts. - **Block public sharing of EBS snapshots (declarative policy)** (AWS): AWS documents that snapshots already publicly shared remain publicly available under block_new_sharing, and EnableSnapshotBlockPublicAccess and DisableSnapshotBlockPublicAccess cannot be used to change the enforced setting. - **Storage SAS tokens should adhere to 7 day maximum validity** (Azure): Existing storage accounts with longer SAS lifetimes are marked non-compliant but not changed. - **Domain restricted sharing** (Google Cloud): Google documents that the organization principal set and Google Workspace ID are not automatically allowed, so they must be listed to let your own principals be granted roles. - **Service perimeter around managed services** (Google Cloud): Google documents that a VM in a VPC network inside the perimeter cannot reach Cloud Storage buckets outside it, and that ingress or egress rules are needed for any cross-perimeter access. - **Allowed destinations for exporting resources** (Google Cloud): Google documents that moves within an organization are not governed by this constraint, and that the move still needs the usual IAM permissions. - **Custom constraint limiting IAM role grants to organization members** (Google Cloud): Google documents that with the ALLOW action every case not explicitly listed in the condition is blocked, so any other member type in an allow policy is refused. - **Principal access boundary limiting the resources principals can access** (Google Cloud): Google documents that a policy only blocks permissions in its enforcement version, so an access attempt using a permission it cannot block is not stopped. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Public exposure: 10 preventive guardrails across three clouds](/risk/public-exposure/) - [Logging and monitoring: 8 preventive guardrails across three clouds](/risk/logging-and-monitoring/) - [Encryption and keys: 13 preventive guardrails across three clouds](/risk/encryption-and-keys/) ## Sources - Deny AWS RAM sharing with external accounts: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-RAM-from-sharing-resources-to-external-accounts.json (read 10 Nov 2026) - Block public sharing of EBS snapshots (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - Storage SAS tokens should adhere to 7 day maximum validity: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/EnforceStorageSasMaxExpiration7Days_Deny.json (read 10 Nov 2026) - Domain restricted sharing: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains (read 10 Nov 2026) - Service perimeter around managed services: https://cloud.google.com/vpc-service-controls/docs/create-service-perimeters (read 10 Nov 2026) - Allowed destinations for exporting resources: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - Custom constraint limiting IAM role grants to organization members: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains (read 10 Nov 2026) - Principal access boundary limiting the resources principals can access: https://cloud.google.com/iam/docs/principal-access-boundary-policies-create (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/data-exfiltration/ --- # Public exposure guardrails close the open door, and existing resources often stay open Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 10 guardrails in the library address public exposure: 4 AWS, 3 Azure, 3 Google Cloud. The first in the list, "Deny disabling block public access for AMIs", denies ec2:DisableImageBlockPublicAccess so Block public access for AMIs cannot be turned off. Each entry gives the policy text, what it blocks, what it is known to break and how to test it first. Guardrails in this class keep resources off the public internet: public blob access, public IPs, open function URLs, unauthenticated identity pools. The library holds 10. Their typical cost is to applications that relied on the open setting. In AWS, requiring IAM authentication on Lambda function URLs blocks any function URL with auth type NONE, and denying unauthenticated Cognito access ends guest credentials. Their typical gap is the resources that already exist: Azure marks an existing network interface with a public IP as noncompliant but does not change it, and Google says its Cloud SQL public IP constraint is not retroactive. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Public exposure | documented | | [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Public exposure | documented | | [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Public exposure | documented | | [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | | [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Public exposure | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny disabling block public access for AMIs** (AWS): The README notes the default for the setting depends on whether the account is new or existing and whether it has public AMIs; the policy has no exception condition. - **Require IAM authentication on Lambda function URLs** (AWS): Function URLs with auth type NONE can no longer be created or updated in the affected accounts. - **Deny changes to S3 Block Public Access at bucket level** (AWS): No principal, including bucket owners, can change bucket-level Block Public Access settings; the README says S3 policies are recommended over an RCP for this control. - **Deny unauthenticated access to Cognito identity pools** (AWS): Applications that rely on guest access to an identity pool can no longer obtain credentials; only authenticated users can. - **Network interfaces should not have public IPs** (Azure): Existing network interfaces that already have a public IP are marked non-compliant but not changed. - **Configure storage accounts to disallow public blob access** (Azure): Modify rewrites the property during create or update instead of rejecting the request, and existing storage accounts change only through a remediation task. - **Storage accounts should disable public network access** (Azure): The documented effect of the setting is that it denies all logins that match IP or virtual network-based firewall rules. - **Enforce public access prevention on Cloud Storage** (Google Cloud): Google documents that it overrides existing public permissions and revokes public access for existing buckets and objects once enabled. - **Restrict public IP access on Cloud SQL instances** (Google Cloud): Google documents that the constraint is not retroactive: instances with existing Public IP access keep working. - **Define allowed external IPs for VM instances** (Google Cloud): Google documents that it does not restrict the use of IPv6 addresses. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Logging and monitoring: 8 preventive guardrails across three clouds](/risk/logging-and-monitoring/) - [Encryption and keys: 13 preventive guardrails across three clouds](/risk/encryption-and-keys/) - [Network perimeter: 7 preventive guardrails across three clouds](/risk/network-perimeter/) ## Sources - Deny disabling block public access for AMIs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Deny-users-from-disabling-block-public-access-on-AMIs.json (read 10 Nov 2026) - Require IAM authentication on Lambda function URLs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-modification-to-Lambda-URL-Config.json (read 10 Nov 2026) - Deny changes to S3 Block Public Access at bucket level: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Prevent-S3-buckets-from-being-made-public-(Bucket-level).json (read 10 Nov 2026) - Deny unauthenticated access to Cognito identity pools: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/Cognito-Deny-identity-pool-unauthenticated-access.json (read 10 Nov 2026) - Network interfaces should not have public IPs: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkPublicIPNic_Deny.json (read 10 Nov 2026) - Configure storage accounts to disallow public blob access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountDisablePublicBlobAccess_Modify.json (read 10 Nov 2026) - Storage accounts should disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StoragePublicNetworkAccess_AuditDeny.json (read 10 Nov 2026) - Enforce public access prevention on Cloud Storage: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/public-exposure/ --- # Logging and monitoring guardrails protect the audit trail, and block the teams that tune it Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 8 guardrails in the library address logging and monitoring: 4 AWS, 2 Azure, 2 Google Cloud. The first in the list, "Deny enabling and disabling AWS Config", denies stopping or deleting AWS Config recorders, delivery channels, rules and aggregators, and putting new configuration, unless the caller is the privileged role. Guardrails in this class protect the records that show what happened: AWS Config, CloudTrail, GuardDuty and VPC flow logs, plus diagnostic settings in Azure. The library holds 8. They break tuning. Denying changes to AWS Config means teams cannot add their own Config rules or recorders outside the privileged role. The GuardDuty entry lists guardduty:ArchiveFindings and guardduty:CreateFilter, so day-to-day findings handling also needs the privileged role. The flow log entry has no exception condition, so cleanup automation is denied too. Azure's deployIfNotExists is not a blocking effect: matching resources are marked noncompliant until a remediation task runs. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Logging and monitoring | documented | | [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Logging and monitoring | documented | | [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | | [Disable Audit Logging exemption](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Logging and monitoring | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny enabling and disabling AWS Config** (AWS): Teams cannot add or change their own Config rules or recorders outside the privileged role; the README notes AWS Control Tower already applies a similar default control. - **Deny modifications to specific CloudTrail trails** (AWS): The resource is a placeholder trail name that must be replaced; the README notes an organization trail already prevents member account users from altering it. - **Deny disabling or modifying Amazon GuardDuty** (AWS): The action list includes guardduty:ArchiveFindings and guardduty:CreateFilter, so day-to-day findings handling also needs the privileged role. - **Deny deleting VPC flow logs** (AWS): The policy has no exception condition, so it applies to every IAM user, role and the root user in the affected member accounts, including cleanup automation that deletes flow logs. - **Deploy diagnostic settings for Key Vault to Event Hub** (Azure): DeployIfNotExists is not a blocking effect: matching resources are marked non-compliant and no action is taken on them until a remediation task runs. - **Application Insights components should block non-Entra ingestion** (Azure): Existing components that accept unauthenticated ingestion are marked non-compliant but not changed. - **Require predefined policies for VPC flow logs** (Google Cloud): Google documents that subnetworks in scope must have flow logs enabled with the required minimum sampling rate, where by default any settings are allowed. - **Disable Audit Logging exemption** (Google Cloud): Google documents that it does not affect audit-logging exemptions that existed before the constraint was enforced. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Encryption and keys: 13 preventive guardrails across three clouds](/risk/encryption-and-keys/) - [Network perimeter: 7 preventive guardrails across three clouds](/risk/network-perimeter/) - [Region and residency: 5 preventive guardrails across three clouds](/risk/region-and-residency/) ## Sources - Deny enabling and disabling AWS Config: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-enabling-and-disabling-AWS-Config.json (read 10 Nov 2026) - Deny modifications to specific CloudTrail trails: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-modifications-to-specific-CloudTrail-trails.json (read 10 Nov 2026) - Deny disabling or modifying Amazon GuardDuty: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-users-from-disabling-Amazon-GuardDuty-or-modifying-its-configuration.json (read 10 Nov 2026) - Deny deleting VPC flow logs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-users-from-deleting-Amazon-VPC-flow-logs.json (read 10 Nov 2026) - Deploy diagnostic settings for Key Vault to Event Hub: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/DiagnosticLog_DINE.json (read 10 Nov 2026) - Application Insights components should block non-Entra ingestion: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Monitoring/ApplicationInsightsComponents_DisableLocalAuth_Deny.json (read 10 Nov 2026) - Require predefined policies for VPC flow logs: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - Disable Audit Logging exemption: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/logging-and-monitoring/ --- # Encryption and key guardrails enforce the setting, and break clients that cannot meet it Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 13 guardrails in the library address encryption and keys: 7 AWS, 3 Azure, 3 Google Cloud. The first in the list, "Deny KMS key deletion", denies scheduling KMS key deletion and deleting aliases, custom key stores and imported key material unless the caller is the privileged role. Guardrails in this class enforce encryption on write and in transit, and protect keys from deletion. The library holds 13, 7 for AWS. They break whatever cannot meet the requirement. Requiring server-side encryption on S3 uploads makes upload requests that omit the header fail for every principal in the affected accounts. Resource control policies that require HTTPS or TLS 1.2 also deny callers outside the organization, including trusted partners, and AWS warns that deny-based RCPs can block unintentionally. A 30-day KMS deletion window fails any shorter schedule, and denying key deletion means routine key cleanup needs the privileged role. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Encryption and keys | documented | | [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Require TLS 1.2 or later for S3](/guardrails/aws-rcp-s3-minimum-tls-1-2/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Deny S3 customer-provided encryption keys (SSE-C)](/guardrails/aws-rcp-deny-s3-sse-c/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Encryption and keys | documented | | [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Encryption and keys | documented | | [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | | [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Encryption and keys | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny KMS key deletion** (AWS): Routine key cleanup, including alias deletion, needs the privileged role in every affected account. - **Enforce a 30-day KMS key deletion window** (AWS): Scheduling key deletion with a pending window under 30 days fails for every principal in the affected accounts, because the policy has no exception condition. - **Deny disabling default EBS encryption** (AWS): The README says to enable encryption by default in your accounts before implementing this policy. - **Require server-side encryption on S3 uploads** (AWS): Upload requests that omit the server-side encryption header fail for every principal in the affected accounts; the policy has no exception condition. - **Require HTTPS for access to resources** (AWS): Callers outside the organization, including trusted partners, are also denied if they connect over an unencrypted connection, because RCPs apply regardless of whether the principal belongs to the organization. - **Require TLS 1.2 or later for S3** (AWS): Clients limited to older TLS versions, including principals outside the organization, are denied; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. - **Deny S3 customer-provided encryption keys (SSE-C)** (AWS): Uploads that use customer-provided keys are denied for all callers; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. - **Storage accounts should have the specified minimum TLS version** (Azure): Storage accounts below the minimum are rejected: deny returns 403 on create and update. - **Key vaults should have deletion protection enabled** (Azure): Purge protection enforces a mandatory retention period: no one in your organization or Microsoft can purge the vault during the soft delete retention period. - **OS and data disks should be encrypted with a customer-managed key** (Azure): You take on key lifecycle work: the definition says you have full control and responsibility for the key lifecycle, including rotation and management. - **Restrict which services may create resources without CMEK** (Google Cloud): Google documents that enforcement is not retroactive and that existing non-CMEK resources must be reconfigured or recreated manually. - **Restrict unencrypted HTTP access to Cloud Storage** (Google Cloud): Google documents that the Cloud Storage XML API allows unencrypted HTTP by default, so clients using it over HTTP are refused once the constraint is enforced. - **Restrict key destruction to disabled key versions** (Google Cloud): Google documents that the constraint applies to both new and existing key versions. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Network perimeter: 7 preventive guardrails across three clouds](/risk/network-perimeter/) - [Region and residency: 5 preventive guardrails across three clouds](/risk/region-and-residency/) - [Organization integrity: 9 preventive guardrails across three clouds](/risk/organization-integrity/) ## Sources - Deny KMS key deletion: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-KMS-key-deletion.json (read 10 Nov 2026) - Enforce a 30-day KMS key deletion window: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Enforce-30-days-for-KMS-deletion.json (read 10 Nov 2026) - Deny disabling default EBS encryption: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-disabling-default-EBS-encryption.json (read 10 Nov 2026) - Require server-side encryption on S3 uploads: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-S3/Prevent-Amazon-S3-unencrypted-object-uploads.json (read 10 Nov 2026) - Require HTTPS for access to resources: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Restrict-resource-access-patterns/Restrict-access-to-only-HTTPS-connections-to-your-resources.json (read 10 Nov 2026) - Require TLS 1.2 or later for S3: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Enforce-TLS-version.json (read 10 Nov 2026) - Deny S3 customer-provided encryption keys (SSE-C): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Deny-SSE-C.json (read 10 Nov 2026) - Storage accounts should have the specified minimum TLS version: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountMinimumTLSVersion_Audit.json (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/encryption-and-keys/ --- # Network perimeter guardrails restrict connectivity, and they cover less than the name suggests Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 7 guardrails in the library address network perimeter: 2 AWS, 3 Azure, 2 Google Cloud. The first in the list, "Prevent creating or expanding public connectivity for VPCs", denies attaching or creating internet gateways, creating VPC peering connections and creating or updating AWS Global Accelerator resources unless the caller is the privileged role. Guardrails in this class restrict public connectivity, forwarding, firewalls and default networks. The library holds 7. They break connectivity someone needed, and they stop less than readers assume. The AWS entry that prevents creating or expanding public VPC connectivity does not block existing direct internet access or access routed through an on-premises network, per its README. AWS's declarative VPC block public access policy stops administrators changing the enforced configuration while attached. Azure's gateway subnet definition blocks a network security group that would make the gateway stop working. Google's skip-default-network constraint means new projects have no default network. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Network perimeter | documented | | [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | Network perimeter | documented | | [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Network perimeter | documented | | [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | | [Restrict VPC peering usage](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Network perimeter | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Prevent creating or expanding public connectivity for VPCs** (AWS): The README states it does not block existing direct internet access or access routed through an on-premises network, and VPC peering and Global Accelerator changes also need the privileged role. - **Block internet gateway ingress for VPCs (declarative policy)** (AWS): While the policy is attached, account administrators cannot change the enforced configuration; AWS lists ModifyVpcBlockPublicAccessOptions, CreateVpcBlockPublicAccessExclusion and ModifyVpcBlockPublicAccessExclusion as blocked. - **Network interfaces should disable IP forwarding** (Azure): Interfaces with IP forwarding enabled are rejected, and the definition says the setting should be reviewed by the network security team. - **Gateway subnets should not have a network security group** (Azure): You cannot filter gateway subnet traffic with an NSG while this applies; the definition states that an NSG on a gateway subnet makes the gateway stop functioning. - **Configure key vaults to enable the firewall** (Azure): Modify changes the property during create or update instead of rejecting the request, and existing vaults change only through a remediation task. - **Skip default network creation** (Google Cloud): Google documents that, by default, a default network and supporting resources are created with each project, so projects created under this constraint have none and need a network created for them. - **Restrict VPC peering usage** (Google Cloud): Google documents that each peering end is required to have peering permission, so both sides must allow the peer. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Region and residency: 5 preventive guardrails across three clouds](/risk/region-and-residency/) - [Organization integrity: 9 preventive guardrails across three clouds](/risk/organization-integrity/) - [AI services: 19 preventive guardrails across three clouds](/risk/ai-services/) ## Sources - Prevent creating or expanding public connectivity for VPCs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-creating-or-expanding-public-connectivity-for-VPCs.json (read 10 Nov 2026) - Block internet gateway ingress for VPCs (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - Network interfaces should disable IP forwarding: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkIPForwardingNic_Deny.json (read 10 Nov 2026) - Gateway subnets should not have a network security group: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkSecurityGroupOnGatewaySubnet_Deny.json (read 10 Nov 2026) - Configure key vaults to enable the firewall: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/FirewallEnabled_Modify.json (read 10 Nov 2026) - Skip default network creation: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - Restrict VPC peering usage: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/network-perimeter/ --- # Region guardrails keep resources where the law says, and break global services that are not exempted Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 5 guardrails in the library address region and residency: 2 AWS, 2 Azure, 1 Google Cloud. The first in the list, "Deny access outside approved Regions", denies every action requested in a Region outside the approved list (eu-central-1 and eu-west-1 in the example), with a NotAction list that exempts global services, unless the caller is the privileged role. Guardrails in this class limit where resources can be created or used. The library holds 5. In AWS, denying access outside approved Regions breaks any global service missing from the exemption list, and the README warns the example may not include all the latest global services or operations. Azure's allowed locations definition returns 403 for deployments to other locations, and excludes resource groups and a few resource types. Google says that allowing or denying a multi-region does not imply the same for its sub-locations. Opt-in Regions can also be locked, so account owners cannot turn them on themselves. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Region and residency | documented | | [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Region and residency | documented | | [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/) | Azure | [Management group assignment](/mechanisms/management-groups/) | Region and residency | documented | | [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Region and residency | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny access outside approved Regions** (AWS): Global services missing from the exemption list break: the README warns the example might not include all of the latest global AWS services or operations. - **Deny enabling or disabling opt-in Regions** (AWS): Account owners cannot turn opt-in Regions on or off themselves; the README restricts these actions to a privileged role. - **Allowed locations for new resources** (Azure): Deployments to other locations return 403, and the definition itself excludes resource groups, Microsoft.AzureActiveDirectory/b2cDirectories and resources that use the 'global' region. - **Allowed locations assigned at a management group** (Azure): Policy and access assigned at a management group cascade to all child subscriptions, and scopes listed in notScopes on the assignment are skipped entirely: they are not evaluated or included in the compliance count. - **Resource location restriction** (Google Cloud): Google documents that allowing or denying a multi-region does not imply the same for its sub-locations, so a deny of the us multi-region still allows resources in us-east1. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Organization integrity: 9 preventive guardrails across three clouds](/risk/organization-integrity/) - [AI services: 19 preventive guardrails across three clouds](/risk/ai-services/) - [Resource and cost: 7 preventive guardrails across three clouds](/risk/resource-and-cost/) ## Sources - Deny access outside approved Regions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region.json (read 10 Nov 2026) - Deny enabling or disabling opt-in Regions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-account-region-enable-and-disable-actions.json (read 10 Nov 2026) - Allowed locations for new resources: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedLocations_Deny.json (read 10 Nov 2026) - Allowed locations assigned at a management group: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) - Resource location restriction: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/region-and-residency/ --- # Organization integrity guardrails protect the structure of the account tree, and apply to everyone Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 9 guardrails in the library address organization integrity: 3 AWS, 3 Azure, 3 Google Cloud. The first in the list, "Deny member accounts leaving the organization", denies organizations:LeaveOrganization so users and roles in affected member accounts cannot remove their account from the AWS organization. Guardrails in this class protect accounts, billing and the resource tree: leaving the organization, closing accounts, changing contacts, deleting resources and locking them. The library holds 9. Several apply to every principal. The leave-organization SCP has no exception condition, so it applies to every user, role and the root user in affected member accounts. Azure's ReadOnly lock on a resource group with a virtual machine prevents starting or restarting it, because those operations use POST. The denyAction effect does not stop removal of child and extension resources when a delete targets the parent. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Organization integrity | documented | | [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/) | AWS | [Resource control policies (RCPs)](/mechanisms/rcp/) | Organization integrity | documented | | [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Organization integrity | documented | | [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/) | Azure | [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/) | Organization integrity | none found | | [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/) | Azure | [Azure resource locks](/mechanisms/resource-lock/) | Organization integrity | none found | | [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | documented | | [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Organization integrity | none found | | [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/) | Google Cloud | [IAM deny policies](/mechanisms/iam-deny/) | Organization integrity | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Deny member accounts leaving the organization** (AWS): The policy has no exception condition, so it applies to every IAM user, role and the root user in the affected member accounts; AWS documents that SCPs do not affect the management account. - **Deny billing, tax and account contact changes** (AWS): Finance or account teams working outside the privileged role cannot change the payment method, tax preferences or contact information; account:CloseAccount is also denied for other roles. - **Restrict access to resources in a suspended OU** (AWS): AWS service principals are exempt by default; the README says to remove the aws:PrincipalIsAWSService condition if you also want to restrict security or logging services. - **Do not allow deletion of resource types** (Azure): The denyAction effect does not stop removal of child and extension resources when a delete targets the parent, and only supports the DELETE action. - **Deployment stack deny settings (Azure RBAC deny assignment)** (Azure): The deny setting applies only to resources the stack defines explicitly, not to resources created implicitly, and you can exclude at most five principals. - **Resource locks (CanNotDelete and ReadOnly)** (Azure): A read-only lock on a resource group that contains a virtual machine prevents all users from starting or restarting it, because those operations use POST requests. - **Restrict resource service usage** (Google Cloud): Google documents that services with a direct dependency on a restricted service are restricted too, for example GKE when Compute Engine is restricted. - **Restrict Shared VPC host project lien removal** (Google Cloud): Google documents that the permission to remove the lien must be granted at the organization level once enforced. - **Deny deleting projects tagged prod** (Google Cloud): Google documents that the denial condition must use the CEL functions for evaluating resource tags, and other functions and operators are not supported. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [AI services: 19 preventive guardrails across three clouds](/risk/ai-services/) - [Resource and cost: 7 preventive guardrails across three clouds](/risk/resource-and-cost/) - [Identity and privilege: 14 preventive guardrails across three clouds](/risk/identity-and-privilege/) ## Sources - Deny member accounts leaving the organization: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-member-accounts-from-leaving-your-AWS-organization.json (read 10 Nov 2026) - Deny billing, tax and account contact changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-billing-modification-action.json (read 10 Nov 2026) - Restrict access to resources in a suspended OU: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/Deny-resource-access-if-the-resource-belongs-to-a-specific-organizational-unit.json (read 10 Nov 2026) - Do not allow deletion of resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/ResourceType_denyActionDelete.json (read 10 Nov 2026) - Deployment stack deny settings (Azure RBAC deny assignment): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - Resource locks (CanNotDelete and ReadOnly): https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) - Restrict resource service usage: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-resources (read 10 Nov 2026) - Restrict Shared VPC host project lien removal: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/organization-integrity/ --- # AI services guardrails limit models, keys and networks, and the library holds 19 of them Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 19 guardrails in the library address ai services: 7 AWS, 7 Azure, 5 Google Cloud. Each entry gives the policy text, what it blocks, what it is known to break and how to test it first. Every score on this page was read from the evidence on 10 Nov 2026 and links to its source. Guardrails in this class govern model access, API keys, public network access and tools for AI services. The library holds 19, the largest class: 7 for AWS, 7 for Azure and 5 for Google Cloud. They break the work of the teams using the services. The AWS Bedrock allow list stops anyone invoking a model not matched by the approved entry, and denying Bedrock API keys stops existing keys and bearer tokens. Azure notes Azure OpenAI Studio requires key access and will not function if it is disabled. Google says enforcing its Grounding with Google Search constraint removes that feature from applications that use it. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Deny IAM and account actions invoked from chat applications](/guardrails/aws-scp-deny-chat-app-iam-actions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | AI services | documented | | [Enforce an Amazon Bedrock guardrail on model inference (declarative policy)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | none found | | [Opt out of AI service content use organization-wide (declarative policy)](/guardrails/aws-declarative-ai-services-opt-out/) | AWS | [Declarative policies](/mechanisms/declarative-policy/) | AI services | documented | | [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure Machine Learning registry deployments restricted to the allowed registry (preview)](/guardrails/azure-policy-ml-registry-models/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Microsoft Foundry agents and toolboxes should not use denied tools (preview)](/guardrails/azure-policy-foundry-denied-tools/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Azure AI Search services should disable public network access](/guardrails/azure-policy-search-disable-public-network/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | AI services | documented | | [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | none found | | [Restrict egress domains for data connectors](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | AI services | documented | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Region deny that allows Bedrock cross-Region inference** (AWS): The placeholders {example-region-1}, {example-region-2} and the inference profile prefix must be replaced, and the README notes model access still needs the separate model approval SCP. - **Allow only approved Amazon Bedrock models** (AWS): Users cannot invoke or subscribe to any model not matched by the approved entry; the placeholders must be replaced with a provider or model identifier before use. - **Deny Amazon Bedrock API keys** (AWS): Existing Bedrock API keys and short-term bearer tokens stop working for every principal in the affected accounts, because the second statement has no exception condition. - **Limit Amazon Bedrock long-term API keys to under 30 days** (AWS): Creating a Bedrock long-term API key with a validity of 30 days or longer fails for every principal in the affected accounts. - **Deny IAM and account actions invoked from chat applications** (AWS): Requests for these IAM and Organizations actions made from chat channels fail for every principal; the same actions made from the console or CLI are not matched by the condition. - **Enforce an Amazon Bedrock guardrail on model inference (declarative policy)** (AWS): The guardrail must be owned by the management account, must have a non-DRAFT version, and needs a resource-based policy allowing organization members to call ApplyGuardrail. - **Opt out of AI service content use organization-wide (declarative policy)** (AWS): AWS documents that opting out deletes the associated historical content shared with AWS before the setting was applied, other than content needed to provide the service. - **Configure Cognitive Services accounts to disable public network access** (Azure): Modify changes the property during create or update instead of rejecting the request, and existing accounts change only through a remediation task. - **Configure Cognitive Services accounts to disable local authentication** (Azure): Modify changes the property during create or update instead of rejecting the request, and existing accounts change only through a remediation task. - **Configure Azure AI Services resources to disable local key access** (Azure): Azure OpenAI Studio requires key access and will not function if key access is disabled, per the definition description. - **Configure Azure Machine Learning workspaces to disable public network access** (Azure): Modify changes the property during create or update instead of rejecting the request, and existing workspaces change only through a remediation task; Microsoft suggests private endpoints to control exposure instead. - **Azure Machine Learning registry deployments restricted to the allowed registry (preview)** (Azure): This definition is a preview, and if no allowed registry name is set the parameter description says all models from all registries are restricted. - **Microsoft Foundry agents and toolboxes should not use denied tools (preview)** (Azure): This definition is a preview, and the same definition supports Audit, so a deny only applies when you set the effect to Deny. - **Azure AI Search services should disable public network access** (Azure): Existing search services that are public are marked non-compliant but not changed; Microsoft suggests private endpoints to limit exposure. - **Restrict public IP access on new Vertex AI Workbench instances** (Google Cloud): Google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. - **Disable root access on new Vertex AI Workbench notebooks and instances** (Google Cloud): Google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. - **Define access to models on Vertex AI** (Google Cloud): Google documents that the Model Garden policy only applies to models in Model Garden, not models registered in the Model Registry, and that a policy cannot exceed 500 allowed and denied values. - **Disable Grounding with Google Search in generative AI APIs** (Google Cloud): Google documents that the feature is enabled by default, so enforcing the constraint removes grounding with Google Search from applications that use it. - **Restrict egress domains for data connectors** (Google Cloud): Google documents that, when enforced, only URLs with an allowed FQDN can be used for data connectors in the covered projects. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Resource and cost: 7 preventive guardrails across three clouds](/risk/resource-and-cost/) - [Identity and privilege: 14 preventive guardrails across three clouds](/risk/identity-and-privilege/) - [Data exfiltration: 8 preventive guardrails across three clouds](/risk/data-exfiltration/) ## Sources - Region deny that allows Bedrock cross-Region inference: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region-with-Bedrock-CRIS.json (read 10 Nov 2026) - Allow only approved Amazon Bedrock models: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-model-invocation-except-approved-models.json (read 10 Nov 2026) - Deny Amazon Bedrock API keys: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys.json (read 10 Nov 2026) - Limit Amazon Bedrock long-term API keys to under 30 days: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys-longer-than-30-days.json (read 10 Nov 2026) - Deny IAM and account actions invoked from chat applications: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Q-Developer/Deny-actions-invoked-through-Amazon-Q-Developer-in-chat-applications-configurations.json (read 10 Nov 2026) - Enforce an Amazon Bedrock guardrail on model inference (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_bedrock_syntax.html (read 10 Nov 2026) - Opt out of AI service content use organization-wide (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out_syntax.html (read 10 Nov 2026) - Configure Cognitive Services accounts to disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DisablePublicNetworkAccess_Modify.json (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/ai-services/ --- # Resource and cost guardrails restrict what can be built, and reject the request with a 403 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. 7 guardrails in the library address resource and cost: 2 AWS, 3 Azure, 2 Google Cloud. The first in the list, "Restrict EC2 launches to approved instance types", denies ec2:RunInstances unless the instance type is in the approved list (t2.micro and t3.micro in the example). Guardrails in this class limit resource types, sizes and purchases. The library holds 7. They are the most direct form of a guardrail: the request is refused. In AWS, a list of approved EC2 instance types means a launch of any other type fails for every principal in the affected accounts, and the example list must be replaced with your own. Azure's allowed virtual machine size definition rejects other sizes with 403 at create or update time, and its allowed resource types definition affects only types that support tags and location unless the mode is changed. Google's load balancer constraint requires every type to be listed explicitly. | Guardrail | Cloud | Mechanism | Risk class | Tested first | | --- | --- | --- | --- | --- | | [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/) | AWS | [Service control policies (SCPs)](/mechanisms/scp/) | Resource and cost | documented | | [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/) | Azure | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Resource and cost | documented | | [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | | [Restrict load balancer creation based on load balancer types](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) | Google Cloud | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Resource and cost | none found | "Tested first" says whether the provider documents a way to see the guardrail's effect before enforcing it. Read 10 Nov 2026. ## What these guardrails break - **Restrict EC2 launches to approved instance types** (AWS): Launches of any instance type not in the list fail for every principal in the affected accounts; the example list must be replaced with the types you approve. - **Restrict AWS Marketplace subscription changes** (AWS): Teams cannot subscribe to or cancel Marketplace products on their own; the README says to restrict these changes to a privileged role. - **Allowed resource types** (Azure): Only resource types that support tags and location are affected unless you duplicate the definition and change its mode to All. - **Not allowed resource types** (Azure): A deny effect returns 403 for any create or update of a listed type, and compliance results are shown only for non-compliant resources. - **Allowed virtual machine size SKUs** (Azure): Requests for VM sizes outside the list are rejected with 403 at create or update time. - **Shared reservations owner projects** (Google Cloud): Google documents the required value format: projects/PROJECT_NUMBER or under:projects/PROJECT_NUMBER. - **Restrict load balancer creation based on load balancer types** (Google Cloud): Google documents that every load balancer type to be allowed or denied must be listed explicitly, from a fixed list of values. **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Identity and privilege: 14 preventive guardrails across three clouds](/risk/identity-and-privilege/) - [Data exfiltration: 8 preventive guardrails across three clouds](/risk/data-exfiltration/) - [Public exposure: 10 preventive guardrails across three clouds](/risk/public-exposure/) ## Sources - Restrict EC2 launches to approved instance types: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Require-Amazon-EC2-instances-to-use-a-specific-type.json (read 10 Nov 2026) - Restrict AWS Marketplace subscription changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-unwanted-cancellation-or-changes-to-AWS-Marketplace-product-subscription.json (read 10 Nov 2026) - Allowed resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedResourceTypes_Deny.json (read 10 Nov 2026) - Not allowed resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/InvalidResourceTypes_Deny.json (read 10 Nov 2026) - Allowed virtual machine size SKUs: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/VMSkusAllowed_Deny.json (read 10 Nov 2026) - Shared reservations owner projects: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - Restrict load balancer creation based on load balancer types: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/risk/resource-and-cost/ --- # Deny member accounts leaving the organization Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny member accounts leaving the organization is an AWS guardrail written in service control policies. Denies organizations:LeaveOrganization so users and roles in affected member accounts cannot remove their account from the AWS organization. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "organizations:LeaveOrganization" ], "Resource": "*" } ] } ``` - **Blocks:** Denies organizations:LeaveOrganization so users and roles in affected member accounts cannot remove their account from the AWS organization. - **Breaks:** The policy has no exception condition, so it applies to every IAM user, role and the root user in the affected member accounts; AWS documents that SCPs do not affect the management account. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-member-accounts-from-leaving-your-AWS-organization.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `organizations:LeaveOrganization`. ## What the source says `organizations:LeaveOrganization` (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Service control policies](/mechanisms/scp/). ## Related guardrails - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity ## Related pages - [Prevent root credentials management in member accounts (AWS Service control policies)](/guardrails/aws-scp-deny-member-root-credentials/) - [Deny the root user except for S3 bucket policy changes (AWS Service control policies)](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) - [Deny critical IAM user actions (AWS Service control policies)](/guardrails/aws-scp-deny-critical-iam-user-actions/) ## Sources - Deny member accounts leaving the organization: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-member-accounts-from-leaving-your-AWS-organization.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-leave-org/ --- # Prevent root credentials management in member accounts Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Prevent root credentials management in member accounts is an AWS guardrail written in service control policies. Denies all actions for the member account root user unless the session was created with sts:AssumeRoot (the aws:AssumedRoot key is not null), so root credentials are managed centrally from the management account. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "*", "Resource": "*", "Condition": { "ArnLike": { "aws:PrincipalArn": [ "arn:aws:iam::*:root" ] }, "Null": { "aws:AssumedRoot": "true" } } } ] } ``` - **Blocks:** Denies all actions for the member account root user unless the session was created with sts:AssumeRoot (the aws:AssumedRoot key is not null), so root credentials are managed centrally from the management account. - **Breaks:** Direct root user sign-in in a member account can no longer perform actions; the repository README notes the SCP restricts the member account root user and has no effect on the management account. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-root-credentials-management-in-member-accounts-in-AWS-Organizations.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `*`. Its conditions test `aws:PrincipalArn`, `aws:AssumedRoot`. ## What the source says `aws:AssumedRoot":"true` (read 10 Nov 2026). On what it breaks: "An SCP restricts permissions for IAM users and roles in member accounts, including the member account's root user. SCPs have no effect on users or roles in the management account." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Deny the root user except for S3 bucket policy changes (AWS Service control policies)](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) - [Deny critical IAM user actions (AWS Service control policies)](/guardrails/aws-scp-deny-critical-iam-user-actions/) - [Deny creating or changing IAM federation providers (AWS Service control policies)](/guardrails/aws-scp-deny-iam-federation-changes/) ## Sources - Prevent root credentials management in member accounts: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-root-credentials-management-in-member-accounts-in-AWS-Organizations.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-member-root-credentials/ --- # Deny the root user except for S3 bucket policy changes Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny the root user except for S3 bucket policy changes is an AWS guardrail written in service control policies. Denies every action by the member account root user except s3:GetBucketPolicy, s3:PutBucketPolicy and s3:DeleteBucketPolicy. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "NotAction": [ "s3:GetBucketPolicy", "s3:PutBucketPolicy", "s3:DeleteBucketPolicy" ], "Resource": "*", "Condition": { "ArnLike": { "aws:PrincipalArn": "arn:aws:iam::*:root" } } } ] } ``` - **Blocks:** Denies every action by the member account root user except s3:GetBucketPolicy, s3:PutBucketPolicy and s3:DeleteBucketPolicy. - **Breaks:** Any other task done as the root user in an affected account fails; the README points to the AWS list of tasks that require root user credentials. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-the-root-user-from-performing-actions-except-S3-bucketpolicy-changes.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 3 actions or patterns: `s3:GetBucketPolicy`, `s3:PutBucketPolicy`, `s3:DeleteBucketPolicy`. Its conditions test `aws:PrincipalArn`. ## What the source says `s3:PutBucketPolicy` (read 10 Nov 2026). On what it breaks: "Restrict use of root user with exceptions for S3 bucket policy changes, if you are frequently locked out of S3 buckets." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Deny critical IAM user actions (AWS Service control policies)](/guardrails/aws-scp-deny-critical-iam-user-actions/) - [Deny creating or changing IAM federation providers (AWS Service control policies)](/guardrails/aws-scp-deny-iam-federation-changes/) - [Prevent creating or expanding public connectivity for VPCs (AWS Service control policies)](/guardrails/aws-scp-prevent-vpc-public-connectivity/) ## Sources - Deny the root user except for S3 bucket policy changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-the-root-user-from-performing-actions-except-S3-bucketpolicy-changes.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-root-except-s3-bucket-policy/ --- # Deny critical IAM user actions Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny critical IAM user actions is an AWS guardrail written in service control policies. Denies creating IAM users, access keys, login profiles and service-specific credentials, and changing the account password policy, unless the caller is a privileged role or carries the IAMUserManagementException tag set to true. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "iam:CreateAccessKey", "iam:CreateUser", "iam:UpdateLoginProfile", "iam:CreateLoginProfile", "iam:DeleteLoginProfile", "iam:UpdateAccountPasswordPolicy", "iam:DeleteAccountPasswordPolicy", "iam:CreateServiceSpecificCredential" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalArn": "arn:aws:iam::*:role/[PRIVILEGED_ROLE]" }, "StringNotLike": { "aws:PrincipalTag/IAMUserManagementException": "true" } } } ] } ``` - **Blocks:** Denies creating IAM users, access keys, login profiles and service-specific credentials, and changing the account password policy, unless the caller is a privileged role or carries the IAMUserManagementException tag set to true. - **Breaks:** Anyone outside the named privileged role or tag exception cannot create or rotate IAM user credentials, and the placeholder [PRIVILEGED_ROLE] must be replaced before use. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-critical-IAM-user-actions.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 8 actions or patterns: `iam:CreateAccessKey`, `iam:CreateUser`, `iam:UpdateLoginProfile`, `iam:CreateLoginProfile`, `iam:DeleteLoginProfile`, `iam:UpdateAccountPasswordPolicy`, `iam:DeleteAccountPasswordPolicy`, `iam:CreateServiceSpecificCredential`. Its conditions test `aws:PrincipalArn`, `aws:PrincipalTag/IAMUserManagementException`. ## What the source says `iam:CreateAccessKey` (read 10 Nov 2026). On what it breaks: "Restrict creation and modification of IAM user profiles, IAM user access keys, service specific credentials such as Amazon Bedrock API keys and account password policy to a privileged role, or principal with the `IAMUserManagementException` Principal tag set to a value of `true`." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Deny creating or changing IAM federation providers (AWS Service control policies)](/guardrails/aws-scp-deny-iam-federation-changes/) - [Prevent creating or expanding public connectivity for VPCs (AWS Service control policies)](/guardrails/aws-scp-prevent-vpc-public-connectivity/) - [Deny billing, tax and account contact changes (AWS Service control policies)](/guardrails/aws-scp-deny-billing-and-contact-changes/) ## Sources - Deny critical IAM user actions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-critical-IAM-user-actions.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-critical-iam-user-actions/ --- # Deny creating or changing IAM federation providers Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny creating or changing IAM federation providers is an AWS guardrail written in service control policies. Denies creating, updating or deleting IAM SAML and OpenID Connect identity providers unless the caller is the privileged role. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "iam:CreateOpenIDConnectProvider", "iam:CreateSAMLProvider", "iam:DeleteOpenIDConnectProvider", "iam:DeleteSAMLProvider", "iam:UpdateOpenIDConnectProviderThumbprint", "iam:UpdateSAMLProvider" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies creating, updating or deleting IAM SAML and OpenID Connect identity providers unless the caller is the privileged role. - **Breaks:** Teams that manage their own identity providers in member accounts are blocked unless they use the privileged role; the README says it is usually paired with the critical IAM user actions policy. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-unwarranted-IAM-federations-creation-modification.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 6 actions or patterns: `iam:CreateOpenIDConnectProvider`, `iam:CreateSAMLProvider`, `iam:DeleteOpenIDConnectProvider`, `iam:DeleteSAMLProvider`, `iam:UpdateOpenIDConnectProviderThumbprint`, `iam:UpdateSAMLProvider`. Its conditions test `aws:PrincipalARN`. ## What the source says `iam:CreateSAMLProvider` (read 10 Nov 2026). On what it breaks: "Restrict the creation of new and modification of existing IAM federation, this policy is usually used in conjunction with" ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Prevent creating or expanding public connectivity for VPCs (AWS Service control policies)](/guardrails/aws-scp-prevent-vpc-public-connectivity/) - [Deny billing, tax and account contact changes (AWS Service control policies)](/guardrails/aws-scp-deny-billing-and-contact-changes/) - [Deny enabling and disabling AWS Config (AWS Service control policies)](/guardrails/aws-scp-deny-config-changes/) ## Sources - Deny creating or changing IAM federation providers: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-unwarranted-IAM-federations-creation-modification.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-iam-federation-changes/ --- # Prevent creating or expanding public connectivity for VPCs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Prevent creating or expanding public connectivity for VPCs is an AWS guardrail written in service control policies. Denies attaching or creating internet gateways, creating VPC peering connections and creating or updating AWS Global Accelerator resources unless the caller is the privileged role. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "ec2:AttachInternetGateway", "ec2:CreateInternetGateway", "ec2:CreateEgressOnlyInternetGateway", "ec2:CreateVpcPeeringConnection", "ec2:AcceptVpcPeeringConnection", "globalaccelerator:CreateAccelerator", "globalaccelerator:CreateCrossAccountAttachment", "globalaccelerator:CreateCustomRoutingAccelerator", "globalaccelerator:CreateCustomRoutingEndpointGroup", "globalaccelerator:CreateCustomRoutingListener", "globalaccelerator:CreateEndpointGroup", "globalaccelerator:CreateListener", "globalaccelerator:UpdateAccelerator", "globalaccelerator:UpdateAcceleratorAttributes", "globalaccelerator:UpdateCrossAccountAttachment", "globalaccelerator:UpdateCustomRoutingAccelerator", "globalaccelerator:UpdateCustomRoutingAcceleratorAttributes", "globalaccelerator:UpdateCustomRoutingListener", "globalaccelerator:UpdateEndpointGroup", "globalaccelerator:UpdateListener" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies attaching or creating internet gateways, creating VPC peering connections and creating or updating AWS Global Accelerator resources unless the caller is the privileged role. - **Breaks:** The README states it does not block existing direct internet access or access routed through an on-premises network, and VPC peering and Global Accelerator changes also need the privileged role. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-creating-or-expanding-public-connectivity-for-VPCs.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 20 actions or patterns: `ec2:AttachInternetGateway`, `ec2:CreateInternetGateway`, `ec2:CreateEgressOnlyInternetGateway`, `ec2:CreateVpcPeeringConnection`, `ec2:AcceptVpcPeeringConnection`, `globalaccelerator:CreateAccelerator`, `globalaccelerator:CreateCrossAccountAttachment`, `globalaccelerator:CreateCustomRoutingAccelerator`, and more. Its conditions test `aws:PrincipalARN`. ## What the source says `ec2:AttachInternetGateway` (read 10 Nov 2026). On what it breaks: "It doesn't block existing direct access or any access that routes through your on-premises network environment." ## Related guardrails - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/): Azure, Azure Policy (built-in definitions), network perimeter - [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/): Azure, Azure Policy (built-in definitions), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter ## Related pages - [Deny billing, tax and account contact changes (AWS Service control policies)](/guardrails/aws-scp-deny-billing-and-contact-changes/) - [Deny enabling and disabling AWS Config (AWS Service control policies)](/guardrails/aws-scp-deny-config-changes/) - [Deny modifications to specific CloudTrail trails (AWS Service control policies)](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) ## Sources - Prevent creating or expanding public connectivity for VPCs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Prevent-creating-or-expanding-public-connectivity-for-VPCs.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-prevent-vpc-public-connectivity/ --- # Deny billing, tax and account contact changes Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny billing, tax and account contact changes is an AWS guardrail written in service control policies. Denies changes to contact information, billing and payment preferences, tax registration and closing the account unless the caller is the privileged role. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "account:PutContactInformation", "account:PutAlternateContact", "account:DeleteAlternateContact", "account:CloseAccount", "billing:UpdateBillingPreferences", "billing:PutContractInformation", "billing:UpdateIAMAccessPreference", "invoicing:PutInvoiceEmailDeliveryPreferences", "payments:DeletePaymentInstrument", "payments:UpdatePaymentPreferences", "tax:BatchPutTaxRegistration", "tax:DeleteTaxRegistration", "tax:PutTaxInheritance" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies changes to contact information, billing and payment preferences, tax registration and closing the account unless the caller is the privileged role. - **Breaks:** Finance or account teams working outside the privileged role cannot change the payment method, tax preferences or contact information; account:CloseAccount is also denied for other roles. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-billing-modification-action.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 13 actions or patterns: `account:PutContactInformation`, `account:PutAlternateContact`, `account:DeleteAlternateContact`, `account:CloseAccount`, `billing:UpdateBillingPreferences`, `billing:PutContractInformation`, `billing:UpdateIAMAccessPreference`, `invoicing:PutInvoiceEmailDeliveryPreferences`, and more. Its conditions test `aws:PrincipalARN`. ## What the source says `account:CloseAccount` (read 10 Nov 2026). On what it breaks: "Restrict IAM principals in accounts from making changes to the payment method and tax preferences, changing contact information." ## Related guardrails - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity ## Related pages - [Deny enabling and disabling AWS Config (AWS Service control policies)](/guardrails/aws-scp-deny-config-changes/) - [Deny modifications to specific CloudTrail trails (AWS Service control policies)](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) - [Deny disabling or modifying Amazon GuardDuty (AWS Service control policies)](/guardrails/aws-scp-deny-guardduty-changes/) ## Sources - Deny billing, tax and account contact changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/Deny-billing-modification-action.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Privileged-access-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-billing-and-contact-changes/ --- # Deny enabling and disabling AWS Config Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny enabling and disabling AWS Config is an AWS guardrail written in service control policies. Denies stopping or deleting AWS Config recorders, delivery channels, rules and aggregators, and putting new configuration, unless the caller is the privileged role. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "config:DeleteConfigurationAggregator", "config:DeleteConfigurationRecorder", "config:DeleteDeliveryChannel", "config:PutConfigRule", "config:DeleteConfigRule", "config:DeleteOrganizationConfigRule", "config:DeleteRetentionConfiguration", "config:PutConfigurationAggregator", "config:PutConfigurationRecorder", "config:PutDeliveryChannel", "config:PutRetentionConfiguration", "config:StopConfigurationRecorder", "config:DeliverConfigSnapshot", "config:DeleteAggregationAuthorization", "config:DeleteEvaluationResults", "config:PutOrganizationConfigRule", "config:PutRemediationConfigurations", "config:DeleteRemediationConfiguration", "config:DeleteServiceLinkedConfigurationRecorder", "config:DeleteRemediationExceptions", "config:DeletePendingAggregationRequest" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies stopping or deleting AWS Config recorders, delivery channels, rules and aggregators, and putting new configuration, unless the caller is the privileged role. - **Breaks:** Teams cannot add or change their own Config rules or recorders outside the privileged role; the README notes AWS Control Tower already applies a similar default control. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-enabling-and-disabling-AWS-Config.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 21 actions or patterns: `config:DeleteConfigurationAggregator`, `config:DeleteConfigurationRecorder`, `config:DeleteDeliveryChannel`, `config:PutConfigRule`, `config:DeleteConfigRule`, `config:DeleteOrganizationConfigRule`, `config:DeleteRetentionConfiguration`, `config:PutConfigurationAggregator`, and more. Its conditions test `aws:PrincipalARN`. ## What the source says `config:StopConfigurationRecorder` (read 10 Nov 2026). On what it breaks: "Restrict enabling/disabling AWS Config to a privileged role." ## Related guardrails - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/): AWS, Service control policies (SCPs), logging and monitoring ## Related pages - [Deny modifications to specific CloudTrail trails (AWS Service control policies)](/guardrails/aws-scp-deny-cloudtrail-trail-changes/) - [Deny disabling or modifying Amazon GuardDuty (AWS Service control policies)](/guardrails/aws-scp-deny-guardduty-changes/) - [Deny deleting VPC flow logs (AWS Service control policies)](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) ## Sources - Deny enabling and disabling AWS Config: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-enabling-and-disabling-AWS-Config.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-config-changes/ --- # Deny modifications to specific CloudTrail trails Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny modifications to specific CloudTrail trails is an AWS guardrail written in service control policies. Denies deleting, updating, stopping or changing event selectors on a named CloudTrail trail unless the caller is the privileged role. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "cloudtrail:DeleteTrail", "cloudtrail:PutEventSelectors", "cloudtrail:StopLogging", "cloudtrail:UpdateTrail", "cloudtrail:CreateTrail" ], "Resource": "arn:aws:cloudtrail:${Region}:${Account}:trail/[CLOUDTRAIL_NAME]", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies deleting, updating, stopping or changing event selectors on a named CloudTrail trail unless the caller is the privileged role. - **Breaks:** The resource is a placeholder trail name that must be replaced; the README notes an organization trail already prevents member account users from altering it. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-modifications-to-specific-CloudTrail-trails.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 5 actions or patterns: `cloudtrail:DeleteTrail`, `cloudtrail:PutEventSelectors`, `cloudtrail:StopLogging`, `cloudtrail:UpdateTrail`, `cloudtrail:CreateTrail`. Its conditions test `aws:PrincipalARN`. ## What the source says `cloudtrail:StopLogging` (read 10 Nov 2026). On what it breaks: "Users in member accounts will not have sufficient permissions to delete the organization trail, turn logging on or off, change what types of events are logged, or otherwise alter the organization trail in any way." ## Related guardrails - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/): AWS, Service control policies (SCPs), logging and monitoring ## Related pages - [Deny disabling or modifying Amazon GuardDuty (AWS Service control policies)](/guardrails/aws-scp-deny-guardduty-changes/) - [Deny deleting VPC flow logs (AWS Service control policies)](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) - [Deny KMS key deletion (AWS Service control policies)](/guardrails/aws-scp-deny-kms-key-deletion/) ## Sources - Deny modifications to specific CloudTrail trails: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-modifications-to-specific-CloudTrail-trails.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-cloudtrail-trail-changes/ --- # Deny disabling or modifying Amazon GuardDuty Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny disabling or modifying Amazon GuardDuty is an AWS guardrail written in service control policies. Denies GuardDuty actions that disable detectors or change filters, IP sets, members and publishing destinations unless the caller is the privileged role. What it breaks: the action list includes guardduty:ArchiveFindings and guardduty:CreateFilter, so day-to-day findings handling also needs the privileged role. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "guardduty:AcceptInvitation", "guardduty:ArchiveFindings", "guardduty:CreateDetector", "guardduty:CreateFilter", "guardduty:CreateIPSet", "guardduty:CreateMembers", "guardduty:CreatePublishingDestination", "guardduty:CreateSampleFindings", "guardduty:CreateThreatIntelSet", "guardduty:DeclineInvitations", "guardduty:DeleteDetector", "guardduty:DeleteFilter", "guardduty:DeleteInvitations", "guardduty:DeleteIPSet", "guardduty:DeleteMembers", "guardduty:DeletePublishingDestination", "guardduty:DeleteThreatIntelSet", "guardduty:DisassociateFromMasterAccount", "guardduty:DisassociateMembers", "guardduty:InviteMembers", "guardduty:StartMonitoringMembers", "guardduty:StopMonitoringMembers", "guardduty:TagResource", "guardduty:UnarchiveFindings", "guardduty:UntagResource", "guardduty:UpdateDetector", "guardduty:UpdateFilter", "guardduty:UpdateFindingsFeedback", "guardduty:UpdateIPSet", "guardduty:UpdatePublishingDestination", "guardduty:UpdateThreatIntelSet" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies GuardDuty actions that disable detectors or change filters, IP sets, members and publishing destinations unless the caller is the privileged role. - **Breaks:** The action list includes guardduty:ArchiveFindings and guardduty:CreateFilter, so day-to-day findings handling also needs the privileged role. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-users-from-disabling-Amazon-GuardDuty-or-modifying-its-configuration.json)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-users-from-disabling-Amazon-GuardDuty-or-modifying-its-configuration.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 31 actions or patterns: `guardduty:AcceptInvitation`, `guardduty:ArchiveFindings`, `guardduty:CreateDetector`, `guardduty:CreateFilter`, `guardduty:CreateIPSet`, `guardduty:CreateMembers`, `guardduty:CreatePublishingDestination`, `guardduty:CreateSampleFindings`, and more. Its conditions test `aws:PrincipalARN`. ## What the source says `guardduty:DeleteDetector` (read 10 Nov 2026). On what it breaks: `guardduty:ArchiveFindings` ## Related guardrails - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/): AWS, Service control policies (SCPs), logging and monitoring ## Related pages - [Deny deleting VPC flow logs (AWS Service control policies)](/guardrails/aws-scp-deny-vpc-flow-log-deletion/) - [Deny KMS key deletion (AWS Service control policies)](/guardrails/aws-scp-deny-kms-key-deletion/) - [Enforce a 30-day KMS key deletion window (AWS Service control policies)](/guardrails/aws-scp-kms-30-day-deletion-window/) ## Sources - Deny disabling or modifying Amazon GuardDuty: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Deny-changes-to-security-services/Deny-users-from-disabling-Amazon-GuardDuty-or-modifying-its-configuration.json (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-guardduty-changes/ --- # Deny deleting VPC flow logs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny deleting VPC flow logs is an AWS guardrail written in service control policies. Denies ec2:DeleteFlowLogs in affected accounts so VPC flow logs cannot be removed. What it breaks: the policy has no exception condition, so it applies to every IAM user, role and the root user in the affected member accounts, including cleanup automation that deletes flow logs. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "ec2:DeleteFlowLogs", "Resource": "*" } ] } ``` - **Blocks:** Denies ec2:DeleteFlowLogs in affected accounts so VPC flow logs cannot be removed. - **Breaks:** The policy has no exception condition, so it applies to every IAM user, role and the root user in the affected member accounts, including cleanup automation that deletes flow logs. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-users-from-deleting-Amazon-VPC-flow-logs.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `ec2:DeleteFlowLogs`. ## What the source says "Action: ec2:DeleteFlowLogs" (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Service control policies](/mechanisms/scp/). ## Related guardrails - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring ## Related pages - [Deny KMS key deletion (AWS Service control policies)](/guardrails/aws-scp-deny-kms-key-deletion/) - [Enforce a 30-day KMS key deletion window (AWS Service control policies)](/guardrails/aws-scp-kms-30-day-deletion-window/) - [Deny disabling default EBS encryption (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) ## Sources - Deny deleting VPC flow logs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-users-from-deleting-Amazon-VPC-flow-logs.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-vpc-flow-log-deletion/ --- # Deny KMS key deletion Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny KMS key deletion is an AWS guardrail written in service control policies. Denies scheduling KMS key deletion and deleting aliases, custom key stores and imported key material unless the caller is the privileged role. What it breaks: routine key cleanup, including alias deletion, needs the privileged role in every affected account. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "kms:ScheduleKeyDeletion", "kms:DeleteAlias", "kms:DeleteCustomKeyStore", "kms:DeleteImportedKeyMaterial" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalArn": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies scheduling KMS key deletion and deleting aliases, custom key stores and imported key material unless the caller is the privileged role. - **Breaks:** Routine key cleanup, including alias deletion, needs the privileged role in every affected account. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-KMS-key-deletion.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 4 actions or patterns: `kms:ScheduleKeyDeletion`, `kms:DeleteAlias`, `kms:DeleteCustomKeyStore`, `kms:DeleteImportedKeyMaterial`. Its conditions test `aws:PrincipalArn`. ## What the source says `kms:ScheduleKeyDeletion` (read 10 Nov 2026). On what it breaks: "Deny the accidental or intentional deletion of a KMS key and only allow specific roles to delete KMS keys." ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Enforce a 30-day KMS key deletion window (AWS Service control policies)](/guardrails/aws-scp-kms-30-day-deletion-window/) - [Deny disabling default EBS encryption (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) - [Require server-side encryption on S3 uploads (AWS Service control policies)](/guardrails/aws-scp-require-s3-server-side-encryption/) ## Sources - Deny KMS key deletion: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-KMS-key-deletion.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-kms-key-deletion/ --- # Enforce a 30-day KMS key deletion window Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Enforce a 30-day KMS key deletion window is an AWS guardrail written in service control policies. Denies kms:ScheduleKeyDeletion when the pending window is shorter than 30 days. What it breaks: scheduling key deletion with a pending window under 30 days fails for every principal in the affected accounts, because the policy has no exception condition. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "kms:ScheduleKeyDeletion", "Resource": "*", "Condition": { "NumericLessThan": { "kms:ScheduleKeyDeletionPendingWindowInDays": "30" } } } ] } ``` - **Blocks:** Denies kms:ScheduleKeyDeletion when the pending window is shorter than 30 days. - **Breaks:** Scheduling key deletion with a pending window under 30 days fails for every principal in the affected accounts, because the policy has no exception condition. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Enforce-30-days-for-KMS-deletion.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `kms:ScheduleKeyDeletion`. Its conditions test `kms:ScheduleKeyDeletionPendingWindowInDays`. ## What the source says "kms:ScheduleKeyDeletionPendingWindowInDays: 30" (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Service control policies](/mechanisms/scp/). ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Deny disabling default EBS encryption (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ebs-default-encryption/) - [Require server-side encryption on S3 uploads (AWS Service control policies)](/guardrails/aws-scp-require-s3-server-side-encryption/) - [Deny access outside approved Regions (AWS Service control policies)](/guardrails/aws-scp-deny-unapproved-regions/) ## Sources - Enforce a 30-day KMS key deletion window: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Enforce-30-days-for-KMS-deletion.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-kms-30-day-deletion-window/ --- # Deny disabling default EBS encryption Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny disabling default EBS encryption is an AWS guardrail written in service control policies. Denies ec2:DisableEbsEncryptionByDefault unless the caller is the privileged role. What it breaks: the README says to enable encryption by default in your accounts before implementing this policy. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "ec2:DisableEbsEncryptionByDefault" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies ec2:DisableEbsEncryptionByDefault unless the caller is the privileged role. - **Breaks:** The README says to enable encryption by default in your accounts before implementing this policy. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-disabling-default-EBS-encryption.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `ec2:DisableEbsEncryptionByDefault`. Its conditions test `aws:PrincipalARN`. ## What the source says `ec2:DisableEbsEncryptionByDefault` (read 10 Nov 2026). On what it breaks: "Note : Enable your accounts for" ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Require server-side encryption on S3 uploads (AWS Service control policies)](/guardrails/aws-scp-require-s3-server-side-encryption/) - [Deny access outside approved Regions (AWS Service control policies)](/guardrails/aws-scp-deny-unapproved-regions/) - [Deny enabling or disabling opt-in Regions (AWS Service control policies)](/guardrails/aws-scp-deny-region-enable-disable/) ## Sources - Deny disabling default EBS encryption: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-disabling-default-EBS-encryption.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-disable-ebs-default-encryption/ --- # Require server-side encryption on S3 uploads Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require server-side encryption on S3 uploads is an AWS guardrail written in service control policies. Denies s3:PutObject when the request does not include the s3:x-amz-server-side-encryption header. What it breaks: upload requests that omit the server-side encryption header fail for every principal in the affected accounts; the policy has no exception condition. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "s3:PutObject", "Resource": "*", "Condition": { "Null": { "s3:x-amz-server-side-encryption": "true" } } } ] } ``` - **Blocks:** Denies s3:PutObject when the request does not include the s3:x-amz-server-side-encryption header. - **Breaks:** Upload requests that omit the server-side encryption header fail for every principal in the affected accounts; the policy has no exception condition. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-S3/Prevent-Amazon-S3-unencrypted-object-uploads.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `s3:PutObject`. Its conditions test `s3:x-amz-server-side-encryption`. ## What the source says "s3:x-amz-server-side-encryption: true" (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Service control policies](/mechanisms/scp/). ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Deny access outside approved Regions (AWS Service control policies)](/guardrails/aws-scp-deny-unapproved-regions/) - [Deny enabling or disabling opt-in Regions (AWS Service control policies)](/guardrails/aws-scp-deny-region-enable-disable/) - [Region deny that allows Bedrock cross-Region inference (AWS Service control policies)](/guardrails/aws-scp-region-deny-with-bedrock-cris/) ## Sources - Require server-side encryption on S3 uploads: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-S3/Prevent-Amazon-S3-unencrypted-object-uploads.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-require-s3-server-side-encryption/ --- # Deny access outside approved Regions Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny access outside approved Regions is an AWS guardrail written in service control policies. Denies every action requested in a Region outside the approved list (eu-central-1 and eu-west-1 in the example), with a NotAction list that exempts global services, unless the caller is the privileged role. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "NotAction": [ "a4b:*", "acm:*", "aws-marketplace-management:*", "aws-marketplace:*", "aws-portal:*", "budgets:*", "ce:*", "chime:*", "cloudfront:*", "config:*", "cur:*", "directconnect:*", "ec2:DescribeRegions", "ec2:DescribeTransitGateways", "ec2:DescribeVpnGateways", "fms:*", "globalaccelerator:*", "health:*", "iam:*", "importexport:*", "kms:*", "mobileanalytics:*", "networkmanager:*", "organizations:*", "pricing:*", "route53:*", "route53domains:*", "route53-recovery-cluster:*", "route53-recovery-control-config:*", "route53-recovery-readiness:*", "s3:GetAccountPublic*", "s3:ListAllMyBuckets", "s3:ListMultiRegionAccessPoints", "s3:PutAccountPublic*", "shield:*", "sts:*", "support:*", "trustedadvisor:*", "waf-regional:*", "waf:*", "wafv2:*", "wellarchitected:*" ], "Resource": "*", "Condition": { "StringNotEquals": { "aws:RequestedRegion": [ "eu-central-1", "eu-west-1" ] }, "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies every action requested in a Region outside the approved list (eu-central-1 and eu-west-1 in the example), with a NotAction list that exempts global services, unless the caller is the privileged role. - **Breaks:** Global services missing from the exemption list break: the README warns the example might not include all of the latest global AWS services or operations. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 42 actions or patterns: `a4b:*`, `acm:*`, `aws-marketplace-management:*`, `aws-marketplace:*`, `aws-portal:*`, `budgets:*`, `ce:*`, `chime:*`, and more. Its conditions test `aws:RequestedRegion`, `aws:PrincipalARN`. ## What the source says `aws:RequestedRegion` (read 10 Nov 2026). On what it breaks: "This example might not include all of the latest global AWS services or operations." ## Related guardrails - [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/): Azure, Azure Policy (built-in definitions), region and residency - [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/): Azure, Management group assignment, region and residency - [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/): Google Cloud, Organization Policy Service (managed and legacy constraints), region and residency - [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/): AWS, Service control policies (SCPs), region and residency ## Related pages - [Deny enabling or disabling opt-in Regions (AWS Service control policies)](/guardrails/aws-scp-deny-region-enable-disable/) - [Region deny that allows Bedrock cross-Region inference (AWS Service control policies)](/guardrails/aws-scp-region-deny-with-bedrock-cris/) - [Allow only approved Amazon Bedrock models (AWS Service control policies)](/guardrails/aws-scp-bedrock-approved-models-only/) ## Sources - Deny access outside approved Regions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-unapproved-regions/ --- # Deny enabling or disabling opt-in Regions Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny enabling or disabling opt-in Regions is an AWS guardrail written in service control policies. Denies account:EnableRegion and account:DisableRegion unless the caller is the privileged role. What it breaks: account owners cannot turn opt-in Regions on or off themselves; the README restricts these actions to a privileged role. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "account:EnableRegion", "account:DisableRegion" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies account:EnableRegion and account:DisableRegion unless the caller is the privileged role. - **Breaks:** Account owners cannot turn opt-in Regions on or off themselves; the README restricts these actions to a privileged role. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-account-region-enable-and-disable-actions.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 2 actions or patterns: `account:EnableRegion`, `account:DisableRegion`. Its conditions test `aws:PrincipalARN`. ## What the source says `account:EnableRegion` (read 10 Nov 2026). On what it breaks: "Restrict enabling or disabling" ## Related guardrails - [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/): Azure, Azure Policy (built-in definitions), region and residency - [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/): Azure, Management group assignment, region and residency - [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/): Google Cloud, Organization Policy Service (managed and legacy constraints), region and residency - [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/): AWS, Service control policies (SCPs), region and residency ## Related pages - [Region deny that allows Bedrock cross-Region inference (AWS Service control policies)](/guardrails/aws-scp-region-deny-with-bedrock-cris/) - [Allow only approved Amazon Bedrock models (AWS Service control policies)](/guardrails/aws-scp-bedrock-approved-models-only/) - [Deny Amazon Bedrock API keys (AWS Service control policies)](/guardrails/aws-scp-deny-bedrock-api-keys/) ## Sources - Deny enabling or disabling opt-in Regions: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-account-region-enable-and-disable-actions.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-region-enable-disable/ --- # Region deny that allows Bedrock cross-Region inference Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Region deny that allows Bedrock cross-Region inference is an AWS guardrail written in service control policies. Extends the Region deny so Amazon Bedrock inference is carved out of the first statement and a second statement denies Bedrock inference outside approved Regions unless the request uses a geographic cross-Region inference profile. ```json { "Version": "2012-10-17", "Statement": [ { "Sid": "DenyUnapprovedRegionsExceptGlobalAndBedrockCRIS", "Effect": "Deny", "NotAction": [ "a4b:*", "acm:*", "aws-marketplace-management:*", "aws-marketplace:*", "aws-portal:*", "bedrock:InvokeModel", "bedrock:InvokeModelWithResponseStream", "bedrock:CreateModelInvocationJob", "budgets:*", "ce:*", "chime:*", "cloudfront:*", "config:*", "cur:*", "directconnect:*", "ec2:DescribeRegions", "ec2:DescribeTransitGateways", "ec2:DescribeVpnGateways", "fms:*", "globalaccelerator:*", "health:*", "iam:*", "importexport:*", "kms:*", "mobileanalytics:*", "networkmanager:*", "organizations:*", "pricing:*", "route53:*", "route53domains:*", "route53-recovery-cluster:*", "route53-recovery-control-config:*", "route53-recovery-readiness:*", "s3:GetAccountPublic*", "s3:ListAllMyBuckets", "s3:ListMultiRegionAccessPoints", "s3:PutAccountPublic*", "shield:*", "sts:*", "support:*", "trustedadvisor:*", "waf-regional:*", "waf:*", "wafv2:*", "wellarchitected:*" ], "Resource": "*", "Condition": { "StringNotEquals": { "aws:RequestedRegion": [ "{example-region-1}", "{example-region-2}" ] }, "ArnNotLike": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } }, { "Sid": "DenyBedrockInferenceOutsideApprovedRegions", "Effect": "Deny", "Action": [ "bedrock:InvokeModel", "bedrock:InvokeModelWithResponseStream", "bedrock:CreateModelInvocationJob" ], "Resource": "*", "Condition": { "StringNotEquals": { "aws:RequestedRegion": [ "{example-region-1}", "{example-region-2}" ] }, "ArnNotLike": { "bedrock:InferenceProfileArn": "arn:aws:bedrock:*:*:inference-profile/{example-inference-profile-geographic-prefix}" } } } ] } ``` - **Blocks:** Extends the Region deny so Amazon Bedrock inference is carved out of the first statement and a second statement denies Bedrock inference outside approved Regions unless the request uses a geographic cross-Region inference profile. - **Breaks:** The placeholders {example-region-1}, {example-region-2} and the inference profile prefix must be replaced, and the README notes model access still needs the separate model approval SCP. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region-with-Bedrock-CRIS.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 45 actions or patterns: `a4b:*`, `acm:*`, `aws-marketplace-management:*`, `aws-marketplace:*`, `aws-portal:*`, `bedrock:InvokeModel`, `bedrock:InvokeModelWithResponseStream`, `bedrock:CreateModelInvocationJob`, and more. Its conditions test `aws:RequestedRegion`, `aws:PrincipalARN`, `bedrock:InferenceProfileArn`. ## What the source says `DenyBedrockInferenceOutsideApprovedRegions` (read 10 Nov 2026). On what it breaks: "To restrict which models can be invoked, combine this with the" ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/): AWS, Service control policies (SCPs), ai services ## Related pages - [Allow only approved Amazon Bedrock models (AWS Service control policies)](/guardrails/aws-scp-bedrock-approved-models-only/) - [Deny Amazon Bedrock API keys (AWS Service control policies)](/guardrails/aws-scp-deny-bedrock-api-keys/) - [Limit Amazon Bedrock long-term API keys to under 30 days (AWS Service control policies)](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) ## Sources - Region deny that allows Bedrock cross-Region inference: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/Deny-access-to-AWS-based-on-the-requested-AWS-region-with-Bedrock-CRIS.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Region-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-region-deny-with-bedrock-cris/ --- # Allow only approved Amazon Bedrock models Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allow only approved Amazon Bedrock models is an AWS guardrail written in service control policies. Denies bedrock:InvokeModel, InvokeModelWithResponseStream and PutFoundationModelEntitlement for any foundation model outside the approved identifier pattern, and covers bedrock-mantle models in a second statement. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Sid": "DenyUsageOfModelsWithBedrock", "Effect": "Deny", "Action": [ "bedrock:InvokeModel", "bedrock:InvokeModelWithResponseStream", "bedrock:PutFoundationModelEntitlement" ], "NotResource": "arn:aws:bedrock:*::foundation-model/.*" }, { "Sid": "DenyUsageOfModelsWithBedrockMantle", "Effect": "Deny", "Action": "bedrock-mantle:*", "Resource": "*", "Condition": { "StringLike": { "bedrock-mantle:Model": "" } } } ] } ``` - **Blocks:** Denies bedrock:InvokeModel, InvokeModelWithResponseStream and PutFoundationModelEntitlement for any foundation model outside the approved identifier pattern, and covers bedrock-mantle models in a second statement. - **Breaks:** Users cannot invoke or subscribe to any model not matched by the approved entry; the placeholders must be replaced with a provider or model identifier before use. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-model-invocation-except-approved-models.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 4 actions or patterns: `bedrock:InvokeModel`, `bedrock:InvokeModelWithResponseStream`, `bedrock:PutFoundationModelEntitlement`, `bedrock-mantle:*`. Its conditions test `bedrock-mantle:Model`. ## What the source says `DenyUsageOfModelsWithBedrock` (read 10 Nov 2026). On what it breaks: "Restrict users and roles from invoking and subscribing to the unapproved Bedrock foundation models." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/): AWS, Service control policies (SCPs), ai services ## Related pages - [Deny Amazon Bedrock API keys (AWS Service control policies)](/guardrails/aws-scp-deny-bedrock-api-keys/) - [Limit Amazon Bedrock long-term API keys to under 30 days (AWS Service control policies)](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) - [Deny IAM and account actions invoked from chat applications (AWS Service control policies)](/guardrails/aws-scp-deny-chat-app-iam-actions/) ## Sources - Allow only approved Amazon Bedrock models: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-model-invocation-except-approved-models.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-bedrock-approved-models-only/ --- # Deny Amazon Bedrock API keys Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny Amazon Bedrock API keys is an AWS guardrail written in service control policies. Denies creating Bedrock service-specific credentials (long-term API keys) and denies bedrock:CallWithBearerToken and bedrock-mantle:CallWithBearerToken. What it breaks: existing Bedrock API keys and short-term bearer tokens stop working for every principal in the affected accounts, because the second statement has no exception condition. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "iam:CreateServiceSpecificCredential", "Resource": "*", "Condition": { "StringEquals": { "iam:ServiceSpecificCredentialServiceName": "bedrock.amazonaws.com" } } }, { "Effect": "Deny", "Action": [ "bedrock:CallWithBearerToken", "bedrock-mantle:CallWithBearerToken" ], "Resource": "*" } ] } ``` - **Blocks:** Denies creating Bedrock service-specific credentials (long-term API keys) and denies bedrock:CallWithBearerToken and bedrock-mantle:CallWithBearerToken. - **Breaks:** Existing Bedrock API keys and short-term bearer tokens stop working for every principal in the affected accounts, because the second statement has no exception condition. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 3 actions or patterns: `iam:CreateServiceSpecificCredential`, `bedrock:CallWithBearerToken`, `bedrock-mantle:CallWithBearerToken`. Its conditions test `iam:ServiceSpecificCredentialServiceName`. ## What the source says `bedrock:CallWithBearerToken` (read 10 Nov 2026). On what it breaks: "denies the usage of Bedrock API keys with the Amazon Bedrock and Amazon Bedrock Powered by AWS Mantle." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/): AWS, Service control policies (SCPs), ai services ## Related pages - [Limit Amazon Bedrock long-term API keys to under 30 days (AWS Service control policies)](/guardrails/aws-scp-bedrock-api-keys-max-30-days/) - [Deny IAM and account actions invoked from chat applications (AWS Service control policies)](/guardrails/aws-scp-deny-chat-app-iam-actions/) - [Restrict EC2 launches to approved instance types (AWS Service control policies)](/guardrails/aws-scp-restrict-ec2-instance-types/) ## Sources - Deny Amazon Bedrock API keys: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-bedrock-api-keys/ --- # Limit Amazon Bedrock long-term API keys to under 30 days Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Limit Amazon Bedrock long-term API keys to under 30 days is an AWS guardrail written in service control policies. Denies iam:CreateServiceSpecificCredential for Bedrock when the credential age is 30 days or more. What it breaks: creating a Bedrock long-term API key with a validity of 30 days or longer fails for every principal in the affected accounts. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "iam:CreateServiceSpecificCredential", "Resource": "*", "Condition": { "StringEquals": { "iam:ServiceSpecificCredentialServiceName": "bedrock.amazonaws.com" }, "NumericGreaterThanEquals": { "iam:ServiceSpecificCredentialAgeDays": "30" } } } ] } ``` - **Blocks:** Denies iam:CreateServiceSpecificCredential for Bedrock when the credential age is 30 days or more. - **Breaks:** Creating a Bedrock long-term API key with a validity of 30 days or longer fails for every principal in the affected accounts. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys-longer-than-30-days.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `iam:CreateServiceSpecificCredential`. Its conditions test `iam:ServiceSpecificCredentialServiceName`. ## What the source says "iam:ServiceSpecificCredentialAgeDays: 30" (read 10 Nov 2026). On what it breaks: "that are valid for longer than 30 days." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services ## Related pages - [Deny IAM and account actions invoked from chat applications (AWS Service control policies)](/guardrails/aws-scp-deny-chat-app-iam-actions/) - [Restrict EC2 launches to approved instance types (AWS Service control policies)](/guardrails/aws-scp-restrict-ec2-instance-types/) - [Restrict AWS Marketplace subscription changes (AWS Service control policies)](/guardrails/aws-scp-restrict-marketplace-subscriptions/) ## Sources - Limit Amazon Bedrock long-term API keys to under 30 days: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Bedrock/Deny-Bedrock-api-keys-longer-than-30-days.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-bedrock-api-keys-max-30-days/ --- # Deny IAM and account actions invoked from chat applications Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny IAM and account actions invoked from chat applications is an AWS guardrail written in service control policies. Denies IAM, sts:AssumeRole, account and Organizations actions when the request comes through an Amazon Q Developer in chat applications configuration (aws:ChatbotSourceArn). Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "iam:AddUserToGroup", "iam:AttachGroupPolicy", "iam:AttachRolePolicy", "iam:AttachUserPolicy", "iam:CreateAccessKey", "iam:CreateLoginProfile", "iam:CreateServiceSpecificCredential", "iam:DeleteRolePermissionsBoundary", "iam:DeleteUserPermissionsBoundary", "iam:DeleteRolePolicy", "iam:DeleteUserPolicy", "iam:DetachRolePolicy", "iam:DetachUserPolicy", "iam:PutGroupPolicy", "iam:PutRolePolicy", "iam:PutUserPolicy", "iam:ResetServiceSpecificCredential", "iam:SetDefaultPolicyVersion", "iam:UpdateAccessKey", "sts:AssumeRole", "account:GetContactInformation", "organizations:*" ], "Resource": "*", "Condition": { "ArnLike": { "aws:ChatbotSourceArn": "arn:aws:chatbot::*:*" } } } ] } ``` - **Blocks:** Denies IAM, sts:AssumeRole, account and Organizations actions when the request comes through an Amazon Q Developer in chat applications configuration (aws:ChatbotSourceArn). - **Breaks:** Requests for these IAM and Organizations actions made from chat channels fail for every principal; the same actions made from the console or CLI are not matched by the condition. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Q-Developer/Deny-actions-invoked-through-Amazon-Q-Developer-in-chat-applications-configurations.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 22 actions or patterns: `iam:AddUserToGroup`, `iam:AttachGroupPolicy`, `iam:AttachRolePolicy`, `iam:AttachUserPolicy`, `iam:CreateAccessKey`, `iam:CreateLoginProfile`, `iam:CreateServiceSpecificCredential`, `iam:DeleteRolePermissionsBoundary`, and more. Its conditions test `aws:ChatbotSourceArn`. ## What the source says `aws:ChatbotSourceArn` (read 10 Nov 2026). On what it breaks: "Restrict IAM and account administration actions invoked through Amazon Q Developer in chat applications configurations." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services ## Related pages - [Restrict EC2 launches to approved instance types (AWS Service control policies)](/guardrails/aws-scp-restrict-ec2-instance-types/) - [Restrict AWS Marketplace subscription changes (AWS Service control policies)](/guardrails/aws-scp-restrict-marketplace-subscriptions/) - [Deny disabling block public access for AMIs (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ami-block-public-access/) ## Sources - Deny IAM and account actions invoked from chat applications: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-Q-Developer/Deny-actions-invoked-through-Amazon-Q-Developer-in-chat-applications-configurations.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-chat-app-iam-actions/ --- # Restrict EC2 launches to approved instance types Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict EC2 launches to approved instance types is an AWS guardrail written in service control policies. Denies ec2:RunInstances unless the instance type is in the approved list (t2.micro and t3.micro in the example). Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "ec2:RunInstances", "Resource": "arn:aws:ec2:*:*:instance/*", "Condition": { "StringNotEquals": { "ec2:InstanceType": [ "t2.micro", "t3.micro" ] } } } ] } ``` - **Blocks:** Denies ec2:RunInstances unless the instance type is in the approved list (t2.micro and t3.micro in the example). - **Breaks:** Launches of any instance type not in the list fail for every principal in the affected accounts; the example list must be replaced with the types you approve. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Require-Amazon-EC2-instances-to-use-a-specific-type.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `ec2:RunInstances`. Its conditions test `ec2:InstanceType`. ## What the source says `ec2:InstanceType` (read 10 Nov 2026). On what it breaks: "Restrict users and roles from launching EC2 instances unless they use only approved instance types." ## Related guardrails - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/): Azure, Azure Policy (built-in definitions), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost ## Related pages - [Restrict AWS Marketplace subscription changes (AWS Service control policies)](/guardrails/aws-scp-restrict-marketplace-subscriptions/) - [Deny disabling block public access for AMIs (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ami-block-public-access/) - [Deny AWS RAM sharing with external accounts (AWS Service control policies)](/guardrails/aws-scp-deny-ram-external-sharing/) ## Sources - Restrict EC2 launches to approved instance types: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Require-Amazon-EC2-instances-to-use-a-specific-type.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-restrict-ec2-instance-types/ --- # Restrict AWS Marketplace subscription changes Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict AWS Marketplace subscription changes is an AWS guardrail written in service control policies. Denies AWS Marketplace subscribe, unsubscribe and private marketplace changes unless the caller is the privileged role. What it breaks: teams cannot subscribe to or cancel Marketplace products on their own; the README says to restrict these changes to a privileged role. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "aws-marketplace:Subscribe", "aws-marketplace:Unsubscribe", "aws-marketplace:CreatePrivateMarketplace", "aws-marketplace:CreatePrivateMarketplaceRequests", "aws-marketplace:AssociateProductsWithPrivateMarketplace", "aws-marketplace:DisassociateProductsFromPrivateMarketplace", "aws-marketplace:UpdatePrivateMarketplaceSettings" ], "Resource": "*", "Condition": { "ArnNotLike": { "aws:PrincipalArn": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies AWS Marketplace subscribe, unsubscribe and private marketplace changes unless the caller is the privileged role. - **Breaks:** Teams cannot subscribe to or cancel Marketplace products on their own; the README says to restrict these changes to a privileged role. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-unwanted-cancellation-or-changes-to-AWS-Marketplace-product-subscription.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 7 actions or patterns: `aws-marketplace:Subscribe`, `aws-marketplace:Unsubscribe`, `aws-marketplace:CreatePrivateMarketplace`, `aws-marketplace:CreatePrivateMarketplaceRequests`, `aws-marketplace:AssociateProductsWithPrivateMarketplace`, `aws-marketplace:DisassociateProductsFromPrivateMarketplace`, `aws-marketplace:UpdatePrivateMarketplaceSettings`. Its conditions test `aws:PrincipalArn`. ## What the source says `aws-marketplace:Subscribe` (read 10 Nov 2026). On what it breaks: "Restrict AWS Marketplace product subscription changes to privileged role" ## Related guardrails - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/): Azure, Azure Policy (built-in definitions), resource and cost - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost ## Related pages - [Deny disabling block public access for AMIs (AWS Service control policies)](/guardrails/aws-scp-deny-disable-ami-block-public-access/) - [Deny AWS RAM sharing with external accounts (AWS Service control policies)](/guardrails/aws-scp-deny-ram-external-sharing/) - [Require IAM authentication on Lambda function URLs (AWS Service control policies)](/guardrails/aws-scp-require-lambda-url-iam-auth/) ## Sources - Restrict AWS Marketplace subscription changes: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/Deny-unwanted-cancellation-or-changes-to-AWS-Marketplace-product-subscription.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Protect-cloud-platform-resource/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-restrict-marketplace-subscriptions/ --- # Deny disabling block public access for AMIs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny disabling block public access for AMIs is an AWS guardrail written in service control policies. Denies ec2:DisableImageBlockPublicAccess so Block public access for AMIs cannot be turned off. What it breaks: the README notes the default for the setting depends on whether the account is new or existing and whether it has public AMIs; the policy has no exception condition. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "ec2:DisableImageBlockPublicAccess", "Resource": "*" } ] } ``` - **Blocks:** Denies ec2:DisableImageBlockPublicAccess so Block public access for AMIs cannot be turned off. - **Breaks:** The README notes the default for the setting depends on whether the account is new or existing and whether it has public AMIs; the policy has no exception condition. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Deny-users-from-disabling-block-public-access-on-AMIs.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `ec2:DisableImageBlockPublicAccess`. ## What the source says `ec2:DisableImageBlockPublicAccess` (read 10 Nov 2026). On what it breaks: "The Block public access for AMIs setting default configuration depends on whether your" ## Related guardrails - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/): AWS, Resource control policies (RCPs), public exposure ## Related pages - [Deny AWS RAM sharing with external accounts (AWS Service control policies)](/guardrails/aws-scp-deny-ram-external-sharing/) - [Require IAM authentication on Lambda function URLs (AWS Service control policies)](/guardrails/aws-scp-require-lambda-url-iam-auth/) - [Require HTTPS for access to resources (AWS Resource control policies)](/guardrails/aws-rcp-require-https/) ## Sources - Deny disabling block public access for AMIs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Service-specific-controls/Amazon-EC2/Deny-users-from-disabling-block-public-access-on-AMIs.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-disable-ami-block-public-access/ --- # Deny AWS RAM sharing with external accounts Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny AWS RAM sharing with external accounts is an AWS guardrail written in service control policies. Denies ram:CreateResourceShare and ram:UpdateResourceShare when the share allows external principals. What it breaks: sharing resources such as VPC subnets with accounts outside the organization fails for every principal in the affected accounts. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "ram:CreateResourceShare", "ram:UpdateResourceShare" ], "Resource": "*", "Condition": { "Bool": { "ram:RequestedAllowsExternalPrincipals": "true" } } } ] } ``` - **Blocks:** Denies ram:CreateResourceShare and ram:UpdateResourceShare when the share allows external principals. - **Breaks:** Sharing resources such as VPC subnets with accounts outside the organization fails for every principal in the affected accounts. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-RAM-from-sharing-resources-to-external-accounts.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 2 actions or patterns: `ram:CreateResourceShare`, `ram:UpdateResourceShare`. Its conditions test `ram:RequestedAllowsExternalPrincipals`. ## What the source says "ram:RequestedAllowsExternalPrincipals: true" (read 10 Nov 2026). On what it breaks: "Deny users or roles from sharing resources with external account principals." ## Related guardrails - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration ## Related pages - [Require IAM authentication on Lambda function URLs (AWS Service control policies)](/guardrails/aws-scp-require-lambda-url-iam-auth/) - [Require HTTPS for access to resources (AWS Resource control policies)](/guardrails/aws-rcp-require-https/) - [Require TLS 1.2 or later for S3 (AWS Resource control policies)](/guardrails/aws-rcp-s3-minimum-tls-1-2/) ## Sources - Deny AWS RAM sharing with external accounts: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-RAM-from-sharing-resources-to-external-accounts.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-deny-ram-external-sharing/ --- # Require IAM authentication on Lambda function URLs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require IAM authentication on Lambda function URLs is an AWS guardrail written in service control policies. Denies creating or updating a Lambda function URL configuration unless its auth type is AWS_IAM. What it breaks: function URLs with auth type NONE can no longer be created or updated in the affected accounts. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": [ "lambda:CreateFunctionUrlConfig", "lambda:UpdateFunctionUrlConfig" ], "Resource": "arn:aws:lambda:*:*:function/*", "Condition": { "StringNotEquals": { "lambda:FunctionUrlAuthType": "AWS_IAM" } } } ] } ``` - **Blocks:** Denies creating or updating a Lambda function URL configuration unless its auth type is AWS_IAM. - **Breaks:** Function URLs with auth type NONE can no longer be created or updated in the affected accounts. ([source](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md)) - **Test before applying:** As for every Service control policies guardrail: [how to test before enforcing](/mechanisms/scp/) - **Mechanism:** [Service control policies (SCPs)](/mechanisms/scp/): IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-modification-to-Lambda-URL-Config.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/LICENSE ## What the policy text names It names 2 actions or patterns: `lambda:CreateFunctionUrlConfig`, `lambda:UpdateFunctionUrlConfig`. Its conditions test `lambda:FunctionUrlAuthType`. ## What the source says "lambda:FunctionUrlAuthType: AWS_IAM" (read 10 Nov 2026). On what it breaks: "Enforce lambda function URLs are called using IAM authentication." ## Related guardrails - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/): AWS, Resource control policies (RCPs), public exposure ## Related pages - [Require HTTPS for access to resources (AWS Resource control policies)](/guardrails/aws-rcp-require-https/) - [Require TLS 1.2 or later for S3 (AWS Resource control policies)](/guardrails/aws-rcp-s3-minimum-tls-1-2/) - [Deny changes to S3 Block Public Access at bucket level (AWS Resource control policies)](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) ## Sources - Require IAM authentication on Lambda function URLs: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/Deny-modification-to-Lambda-URL-Config.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/service-control-policy-examples/main/Sensitive-data-protection/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-scp-require-lambda-url-iam-auth/ --- # Require HTTPS for access to resources Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require HTTPS for access to resources is an AWS guardrail written in resource control policies. Denies requests to resources in the listed services (S3, STS, KMS, SQS, Secrets Manager, Cognito, CloudWatch Logs, DynamoDB, ECR, OpenSearch Serverless) when aws:SecureTransport is false. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Principal": "*", "Action": [ "s3:*", "sts:*", "kms:*", "sqs:*", "secretsmanager:*", "cognito-identity:*", "cognito-idp:*", "logs:*", "dynamodb:*", "ecr:*", "aoss:*" ], "Resource": "*", "Condition": { "BoolIfExists": { "aws:SecureTransport": "false" } } } ] } ``` - **Blocks:** Denies requests to resources in the listed services (S3, STS, KMS, SQS, Secrets Manager, Cognito, CloudWatch Logs, DynamoDB, ECR, OpenSearch Serverless) when aws:SecureTransport is false. - **Breaks:** Callers outside the organization, including trusted partners, are also denied if they connect over an unencrypted connection, because RCPs apply regardless of whether the principal belongs to the organization. ([source](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Restrict-resource-access-patterns/README.md)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Restrict-resource-access-patterns/Restrict-access-to-only-HTTPS-connections-to-your-resources.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 11 actions or patterns: `s3:*`, `sts:*`, `kms:*`, `sqs:*`, `secretsmanager:*`, `cognito-identity:*`, `cognito-idp:*`, `logs:*`, and more. Its conditions test `aws:SecureTransport`. ## What the source says "aws:SecureTransport: false" (read 10 Nov 2026). On what it breaks: "This can help you prevent potential attackers from manipulating network traffic and trusted partners from accessing your resources over an unencrypted connection." ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Require TLS 1.2 or later for S3 (AWS Resource control policies)](/guardrails/aws-rcp-s3-minimum-tls-1-2/) - [Deny changes to S3 Block Public Access at bucket level (AWS Resource control policies)](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) - [Deny unauthenticated access to Cognito identity pools (AWS Resource control policies)](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) ## Sources - Require HTTPS for access to resources: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Restrict-resource-access-patterns/Restrict-access-to-only-HTTPS-connections-to-your-resources.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Restrict-resource-access-patterns/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-require-https/ --- # Require TLS 1.2 or later for S3 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require TLS 1.2 or later for S3 is an AWS guardrail written in resource control policies. Denies S3 requests that use a TLS version lower than 1.2. What it breaks: clients limited to older TLS versions, including principals outside the organization, are denied; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Principal": "*", "Action": "s3:*", "Resource": "*", "Condition": { "NumericLessThan": { "s3:TlsVersion": [ "1.2" ] } } } ] } ``` - **Blocks:** Denies S3 requests that use a TLS version lower than 1.2. - **Breaks:** Clients limited to older TLS versions, including principals outside the organization, are denied; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Enforce-TLS-version.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `s3:*`. Its conditions test `s3:TlsVersion`. ## What the source says `s3:TlsVersion` (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Resource control policies](/mechanisms/rcp/). ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Deny changes to S3 Block Public Access at bucket level (AWS Resource control policies)](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/) - [Deny unauthenticated access to Cognito identity pools (AWS Resource control policies)](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) - [Deny S3 customer-provided encryption keys (SSE-C) (AWS Resource control policies)](/guardrails/aws-rcp-deny-s3-sse-c/) ## Sources - Require TLS 1.2 or later for S3: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Enforce-TLS-version.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-s3-minimum-tls-1-2/ --- # Deny changes to S3 Block Public Access at bucket level Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny changes to S3 Block Public Access at bucket level is an AWS guardrail written in resource control policies. Denies s3:PutBucketPublicAccessBlock on buckets in the affected accounts so bucket-level Block Public Access settings cannot be modified. Copied from a licensed open source set and read on 10 Nov 2026. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Principal": "*", "Action": "s3:PutBucketPublicAccessBlock", "Resource": "*" } ] } ``` - **Blocks:** Denies s3:PutBucketPublicAccessBlock on buckets in the affected accounts so bucket-level Block Public Access settings cannot be modified. - **Breaks:** No principal, including bucket owners, can change bucket-level Block Public Access settings; the README says S3 policies are recommended over an RCP for this control. ([source](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Prevent-S3-buckets-from-being-made-public-(Bucket-level).json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `s3:PutBucketPublicAccessBlock`. ## What the source says `s3:PutBucketPublicAccessBlock` (read 10 Nov 2026). On what it breaks: "S3 policies is recommended." ## Related guardrails - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/): AWS, Resource control policies (RCPs), public exposure ## Related pages - [Deny unauthenticated access to Cognito identity pools (AWS Resource control policies)](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/) - [Deny S3 customer-provided encryption keys (SSE-C) (AWS Resource control policies)](/guardrails/aws-rcp-deny-s3-sse-c/) - [Restrict access to resources in a suspended OU (AWS Resource control policies)](/guardrails/aws-rcp-restrict-suspended-ou-resources/) ## Sources - Deny changes to S3 Block Public Access at bucket level: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Prevent-S3-buckets-from-being-made-public-(Bucket-level).json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-deny-bucket-public-access-block-changes/ --- # Deny unauthenticated access to Cognito identity pools Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny unauthenticated access to Cognito identity pools is an AWS guardrail written in resource control policies. Denies GetId, GetCredentialsForIdentity and GetOpenIdToken for unauthenticated (guest) identities in Cognito identity pools. What it breaks: applications that rely on guest access to an identity pool can no longer obtain credentials; only authenticated users can. ```json { "Version": "2012-10-17", "Statement": [ { "Sid": "DenyUnauthenticatedIdentityPoolAccess", "Effect": "Deny", "Principal": "*", "Action": [ "cognito-identity:GetId", "cognito-identity:GetCredentialsForIdentity", "cognito-identity:GetOpenIdToken" ], "Resource": "*", "Condition": { "StringLike": { "cognito-identity-unauth:IdentityPoolArn": "*" } } } ] } ``` - **Blocks:** Denies GetId, GetCredentialsForIdentity and GetOpenIdToken for unauthenticated (guest) identities in Cognito identity pools. - **Breaks:** Applications that rely on guest access to an identity pool can no longer obtain credentials; only authenticated users can. ([source](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/README.md)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/Cognito-Deny-identity-pool-unauthenticated-access.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 3 actions or patterns: `cognito-identity:GetId`, `cognito-identity:GetCredentialsForIdentity`, `cognito-identity:GetOpenIdToken`. Its conditions test `cognito-identity-unauth:IdentityPoolArn`. ## What the source says `cognito-identity-unauth:IdentityPoolArn` (read 10 Nov 2026). On what it breaks: "ensuring only authenticated users can obtain credentials." ## Related guardrails - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure ## Related pages - [Deny S3 customer-provided encryption keys (SSE-C) (AWS Resource control policies)](/guardrails/aws-rcp-deny-s3-sse-c/) - [Restrict access to resources in a suspended OU (AWS Resource control policies)](/guardrails/aws-rcp-restrict-suspended-ou-resources/) - [Block internet gateway ingress for VPCs (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-vpc-block-public-access/) ## Sources - Deny unauthenticated access to Cognito identity pools: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/Cognito-Deny-identity-pool-unauthenticated-access.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-deny-cognito-unauthenticated-access/ --- # Deny S3 customer-provided encryption keys (SSE-C) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny S3 customer-provided encryption keys (SSE-C) is an AWS guardrail written in resource control policies. Denies s3:PutObject when the request carries the SSE-C customer algorithm header. What it breaks: uploads that use customer-provided keys are denied for all callers; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. ```json { "Version": "2012-10-17", "Statement": [ { "Sid": "RestrictSSECObjectUploads", "Effect": "Deny", "Principal": "*", "Action": "s3:PutObject", "Resource": "*", "Condition": { "Null": { "s3:x-amz-server-side-encryption-customer-algorithm": "false" } } } ] } ``` - **Blocks:** Denies s3:PutObject when the request carries the SSE-C customer algorithm header. - **Breaks:** Uploads that use customer-provided keys are denied for all callers; AWS warns deny-based RCPs can unintentionally block use of AWS services without needed exceptions. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Deny-SSE-C.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 1 action or pattern: `s3:PutObject`. Its conditions test `s3:x-amz-server-side-encryption-customer-algorithm`. ## What the source says `s3:x-amz-server-side-encryption-customer-algorithm` (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Resource control policies](/mechanisms/rcp/). ## Related guardrails - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys ## Related pages - [Restrict access to resources in a suspended OU (AWS Resource control policies)](/guardrails/aws-rcp-restrict-suspended-ou-resources/) - [Block internet gateway ingress for VPCs (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-vpc-block-public-access/) - [Block public sharing of EBS snapshots (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-snapshot-block-public-access/) ## Sources - Deny S3 customer-provided encryption keys (SSE-C): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Service-specific-controls/S3-Deny-SSE-C.json (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-deny-s3-sse-c/ --- # Restrict access to resources in a suspended OU Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict access to resources in a suspended OU is an AWS guardrail written in resource control policies. Denies a list of actions on resources in the accounts under the OU the policy is attached to, unless the caller is the privileged role or an AWS service principal. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Principal": "*", "Action": [ "s3:*", "sqs:*", "kms:*", "secretsmanager:*", "sts:AssumeRole", "sts:DecodeAuthorizationMessage", "sts:GetAccessKeyInfo", "sts:GetFederationToken", "sts:GetServiceBearerToken", "sts:GetSessionToken", "sts:SetContext", "cognito-identity:*", "cognito-idp:*", "logs:*", "dynamodb:*", "ecr:*", "aoss:*" ], "Resource": "*", "Condition": { "BoolIfExists": { "aws:PrincipalIsAWSService": "false" }, "ArnNotLikeIfExists": { "aws:PrincipalARN": "arn:aws:iam::${Account}:role/[PRIVILEGED_ROLE]" } } } ] } ``` - **Blocks:** Denies a list of actions on resources in the accounts under the OU the policy is attached to, unless the caller is the privileged role or an AWS service principal. - **Breaks:** AWS service principals are exempt by default; the README says to remove the aws:PrincipalIsAWSService condition if you also want to restrict security or logging services. ([source](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/README.md)) - **Test before applying:** As for every Resource control policies guardrail: [how to test before enforcing](/mechanisms/rcp/) - **Mechanism:** [Resource control policies (RCPs)](/mechanisms/rcp/): Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. - **Taken from:** [open source set](https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/Deny-resource-access-if-the-resource-belongs-to-a-specific-organizational-unit.json), MIT-0 (README states MIT-0; LICENSE file fetched, MIT-0 text, Copyright Amazon.com, Inc.): https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/LICENSE ## What the policy text names It names 17 actions or patterns: `s3:*`, `sqs:*`, `kms:*`, `secretsmanager:*`, `sts:AssumeRole`, `sts:DecodeAuthorizationMessage`, `sts:GetAccessKeyInfo`, `sts:GetFederationToken`, and more. Its conditions test `aws:PrincipalIsAWSService`. ## What the source says "aws:PrincipalIsAWSService: false" (read 10 Nov 2026). On what it breaks: "Note: If you would like to restrict AWS services such as security or logging services, you will need to remove the 'aws:PrincipalIsAWSService': 'false' condition." ## Related guardrails - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity ## Related pages - [Block internet gateway ingress for VPCs (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-vpc-block-public-access/) - [Block public sharing of EBS snapshots (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-snapshot-block-public-access/) - [Require IMDSv2 for new EC2 instances (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-require-imdsv2/) ## Sources - Restrict access to resources in a suspended OU: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/Deny-resource-access-if-the-resource-belongs-to-a-specific-organizational-unit.json (read 10 Nov 2026) - What it breaks: https://raw.githubusercontent.com/aws-samples/resource-control-policy-examples/main/Establish-intra-organization-boundaries/README.md (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps_examples.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-rcp-restrict-suspended-ou-resources/ --- # Block internet gateway ingress for VPCs (declarative policy) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Block internet gateway ingress for VPCs (declarative policy) is an AWS guardrail written in declarative policies. Sets VPC Block Public Access to block_ingress for the accounts in scope, so resources in VPCs and subnets cannot receive traffic through internet gateways. ```json { "ec2_attributes": { "vpc_block_public_access": { "internet_gateway_block": { "mode": { "@@assign": "block_ingress" }, "exclusions_allowed": { "@@assign": "enabled" } } } } } ``` - **Blocks:** Sets VPC Block Public Access to block_ingress for the accounts in scope, so resources in VPCs and subnets cannot receive traffic through internet gateways. - **Breaks:** While the policy is attached, account administrators cannot change the enforced configuration; AWS lists ModifyVpcBlockPublicAccessOptions, CreateVpcBlockPublicAccessExclusion and ModifyVpcBlockPublicAccessExclusion as blocked. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html)) - **Test before applying:** As for every Declarative policies guardrail: [how to test before enforcing](/mechanisms/declarative-policy/) - **Mechanism:** [Declarative policies](/mechanisms/declarative-policy/): The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html), provider documentation ## What the source says "vpc_block_public_access: {" (read 10 Nov 2026). On what it breaks: "Account administrators will not be able to modify the value of the service attribute at the individual account level." ## Related guardrails - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/): Azure, Azure Policy (built-in definitions), network perimeter - [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/): Azure, Azure Policy (built-in definitions), network perimeter - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter ## Related pages - [Block public sharing of EBS snapshots (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-snapshot-block-public-access/) - [Require IMDSv2 for new EC2 instances (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-require-imdsv2/) - [Enforce an Amazon Bedrock guardrail on model inference (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) ## Sources - Block internet gateway ingress for VPCs (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-declarative-vpc-block-public-access/ --- # Block public sharing of EBS snapshots (declarative policy) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Block public sharing of EBS snapshots (declarative policy) is an AWS guardrail written in declarative policies. Sets Snapshot Block Public Access to block_new_sharing so new public sharing of EBS snapshots is blocked in the accounts in scope. AWS documents that the account status report shows the current status of the attributes a declarative policy supports for the accounts in scope. ```json { "ec2_attributes": { "snapshot_block_public_access": { "state": { "@@assign": "block_new_sharing" } } } } ``` - **Blocks:** Sets Snapshot Block Public Access to block_new_sharing so new public sharing of EBS snapshots is blocked in the accounts in scope. - **Breaks:** AWS documents that snapshots already publicly shared remain publicly available under block_new_sharing, and EnableSnapshotBlockPublicAccess and DisableSnapshotBlockPublicAccess cannot be used to change the enforced setting. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html)) - **Test before applying:** As for every Declarative policies guardrail: [how to test before enforcing](/mechanisms/declarative-policy/) - **Mechanism:** [Declarative policies](/mechanisms/declarative-policy/): The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html), provider documentation ## What the source says "snapshot_block_public_access: {" (read 10 Nov 2026). On what it breaks: "Blocks new public sharing of snapshots. Snapshots that were already publicly shared remain publicly available." ## Related guardrails - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration ## Related pages - [Require IMDSv2 for new EC2 instances (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-require-imdsv2/) - [Enforce an Amazon Bedrock guardrail on model inference (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) - [Opt out of AI service content use organization-wide (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-ai-services-opt-out/) ## Sources - Block public sharing of EBS snapshots (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-declarative-snapshot-block-public-access/ --- # Require IMDSv2 for new EC2 instances (declarative policy) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require IMDSv2 for new EC2 instances (declarative policy) is an AWS guardrail written in declarative policies. Sets instance metadata defaults to http_tokens required and enforced, so new EC2 launches use IMDSv2 token-backed sessions. AWS documents that the account status report shows the current status of the attributes a declarative policy supports for the accounts in scope. ```json { "ec2_attributes": { "instance_metadata_defaults": { "http_tokens": { "@@assign": "required" }, "http_put_response_hop_limit": { "@@assign": "4" }, "http_endpoint": { "@@assign": "enabled" }, "instance_metadata_tags": { "@@assign": "enabled" }, "http_tokens_enforced": { "@@assign": "enabled" } } } } ``` - **Blocks:** Sets instance metadata defaults to http_tokens required and enforced, so new EC2 launches use IMDSv2 token-backed sessions. - **Breaks:** AWS says to make sure no instances are making IMDSv1 calls before setting http_tokens to required, and recommends a hop limit of at least 2 when it is required. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html)) - **Test before applying:** As for every Declarative policies guardrail: [how to test before enforcing](/mechanisms/declarative-policy/) - **Mechanism:** [Declarative policies](/mechanisms/declarative-policy/): The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html), provider documentation ## What the source says "instance_metadata_defaults: {" (read 10 Nov 2026). On what it breaks: "Before setting http_tokens to required (IMDSv2 must be used), make sure that none of your instances are making IMDSv1 calls." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Enforce an Amazon Bedrock guardrail on model inference (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/) - [Opt out of AI service content use organization-wide (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-ai-services-opt-out/) - [Permissions boundary limiting a user to S3, CloudWatch and EC2 (AWS IAM permissions boundaries)](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) ## Sources - Require IMDSv2 for new EC2 instances (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-declarative-require-imdsv2/ --- # Enforce an Amazon Bedrock guardrail on model inference (declarative policy) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Enforce an Amazon Bedrock guardrail on model inference (declarative policy) is an AWS guardrail written in declarative policies. References a versioned Amazon Bedrock Guardrail from the management account and applies it to inference calls in a Region for all models except the excluded ones. ```json { "bedrock": { "guardrail_inference": { "us-east-1": { "config_1": { "identifier": { "@@assign": "arn:aws:bedrock:us-east-1:123456789012:guardrail/hu1dlsv9wy1d:1" }, "selective_content_guarding": { "system": { "@@assign": "selective" }, "messages": { "@@assign": "comprehensive" } }, "model_enforcement": { "included_models": { "@@assign": [ "ALL" ] }, "excluded_models": { "@@assign": [ "amazon.titan-embed-text-v2:0", "cohere.embed-english-v3" ] } } } } } } } ``` - **Blocks:** References a versioned Amazon Bedrock Guardrail from the management account and applies it to inference calls in a Region for all models except the excluded ones. - **Breaks:** The guardrail must be owned by the management account, must have a non-DRAFT version, and needs a resource-based policy allowing organization members to call ApplyGuardrail. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_bedrock_syntax.html)) - **Test before applying:** As for every Declarative policies guardrail: [how to test before enforcing](/mechanisms/declarative-policy/) - **Mechanism:** [Declarative policies](/mechanisms/declarative-policy/): The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_bedrock_syntax.html), provider documentation ## What the source says "guardrail_inference: {" (read 10 Nov 2026). On what it breaks: "The Guardrail must be owned by the Management account. You cannot create a policy using a Guardrail from another account." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services ## Related pages - [Opt out of AI service content use organization-wide (declarative policy) (AWS Declarative policies)](/guardrails/aws-declarative-ai-services-opt-out/) - [Permissions boundary limiting a user to S3, CloudWatch and EC2 (AWS IAM permissions boundaries)](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) - [Permissions boundary for delegated IAM user creation (AWS IAM permissions boundaries)](/guardrails/aws-pb-delegated-user-creation-boundary/) ## Sources - Enforce an Amazon Bedrock guardrail on model inference (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_bedrock_syntax.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-declarative-bedrock-guardrail-enforcement/ --- # Opt out of AI service content use organization-wide (declarative policy) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Opt out of AI service content use organization-wide (declarative policy) is an AWS guardrail written in declarative policies. Opts all AI services out of using and storing customer content for service improvement, with comprehend locked so child policies cannot override it. ```json { "services": { "default": { "opt_out_policy": { "@@assign": "optOut" } }, "comprehend": { "opt_out_policy": { "@@operators_allowed_for_child_policies": [ "@@none" ], "@@assign": "optOut" } }, "rekognition": { "opt_out_policy": { "@@assign": "optIn" } } } } ``` - **Blocks:** Opts all AI services out of using and storing customer content for service improvement, with comprehend locked so child policies cannot override it. - **Breaks:** AWS documents that opting out deletes the associated historical content shared with AWS before the setting was applied, other than content needed to provide the service. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out.html)) - **Test before applying:** AWS documents that you can query the effective policy applicable to each account to see the effects of your setting choices. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out.html)) - **Mechanism:** [Declarative policies](/mechanisms/declarative-policy/): The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out_syntax.html), provider documentation ## What the source says "opt_out_policy: {" (read 10 Nov 2026). On what it breaks: "When you opt out of content use by an AWS AI service, that service deletes all of the associated historical content that was shared with AWS before you set the option." ## Related guardrails - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services ## Related pages - [Permissions boundary limiting a user to S3, CloudWatch and EC2 (AWS IAM permissions boundaries)](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/) - [Permissions boundary for delegated IAM user creation (AWS IAM permissions boundaries)](/guardrails/aws-pb-delegated-user-creation-boundary/) - [Allowed locations for new resources (Azure Azure Policy)](/guardrails/azure-policy-allowed-locations/) ## Sources - Opt out of AI service content use organization-wide (declarative policy): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out_syntax.html (read 10 Nov 2026) - What it breaks: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ai-opt-out.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-declarative-ai-services-opt-out/ --- # Permissions boundary limiting a user to S3, CloudWatch and EC2 Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Permissions boundary limiting a user to S3, CloudWatch and EC2 is an AWS guardrail written in iAM permissions boundaries. Sets the maximum permissions of an IAM user to s3:*, cloudwatch:* and ec2:*, so the user can never act in other services such as IAM even if an identity policy allows it. ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "s3:*", "cloudwatch:*", "ec2:*" ], "Resource": "*" } ] } ``` - **Blocks:** Sets the maximum permissions of an IAM user to s3:*, cloudwatch:* and ec2:*, so the user can never act in other services such as IAM even if an identity policy allows it. - **Breaks:** AWS gives the example of an iam:CreateUser permission policy failing because the boundary does not allow the action. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html)) - **Test before applying:** AWS documents that the IAM policy simulator can test IAM permissions boundaries, one boundary at a time. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) - **Mechanism:** [IAM permissions boundaries](/mechanisms/permissions-boundary/): A single IAM user or role it is attached to; it does not apply to other principals or to resources. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html), provider documentation ## What the policy text names It names 3 actions or patterns: `s3:*`, `cloudwatch:*`, `ec2:*`. ## What the source says "s3:*, cloudwatch:*, ec2:*" (read 10 Nov 2026). On what it breaks: "It fails because the permissions boundary does not allow the iam:CreateUser operation." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Permissions boundary for delegated IAM user creation (AWS IAM permissions boundaries)](/guardrails/aws-pb-delegated-user-creation-boundary/) - [Allowed locations for new resources (Azure Azure Policy)](/guardrails/azure-policy-allowed-locations/) - [Allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-allowed-resource-types/) ## Sources - Permissions boundary limiting a user to S3, CloudWatch and EC2: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/ --- # Permissions boundary for delegated IAM user creation Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Permissions boundary for delegated IAM user creation is an AWS guardrail written in iAM permissions boundaries. Lets a delegated administrator create or change IAM users only when the XCompanyBoundaries boundary is applied, and denies editing the boundary policies or deleting a user's boundary. ```json { "Version": "2012-10-17", "Statement": [ { "Sid": "CreateOrChangeOnlyWithBoundary", "Effect": "Allow", "Action": [ "iam:AttachUserPolicy", "iam:CreateUser", "iam:DeleteUserPolicy", "iam:DetachUserPolicy", "iam:PutUserPermissionsBoundary", "iam:PutUserPolicy" ], "Resource": "*", "Condition": { "StringEquals": { "iam:PermissionsBoundary": "arn:aws:iam::123456789012:policy/XCompanyBoundaries" } } }, { "Sid": "CloudWatchAndOtherIAMTasks", "Effect": "Allow", "Action": [ "cloudwatch:*", "iam:CreateAccessKey", "iam:CreateGroup", "iam:CreateLoginProfile", "iam:CreatePolicy", "iam:DeleteGroup", "iam:DeletePolicy", "iam:DeletePolicyVersion", "iam:DeleteUser", "iam:GetAccountPasswordPolicy", "iam:GetGroup", "iam:GetLoginProfile", "iam:GetPolicy", "iam:GetPolicyVersion", "iam:GetRolePolicy", "iam:GetUser", "iam:GetUserPolicy", "iam:ListAccessKeys", "iam:ListAttachedRolePolicies", "iam:ListAttachedUserPolicies", "iam:ListEntitiesForPolicy", "iam:ListGroups", "iam:ListGroupsForUser", "iam:ListMFADevices", "iam:ListPolicies", "iam:ListPolicyVersions", "iam:ListRolePolicies", "iam:ListSSHPublicKeys", "iam:ListServiceSpecificCredentials", "iam:ListSigningCertificates", "iam:ListUserPolicies", "iam:ListUsers", "iam:SetDefaultPolicyVersion", "iam:SimulateCustomPolicy", "iam:SimulatePrincipalPolicy", "iam:UpdateGroup", "iam:UpdateLoginProfile", "iam:UpdateUser" ], "NotResource": "arn:aws:iam::123456789012:user/Maria" }, { "Sid": "NoBoundaryPolicyEdit", "Effect": "Deny", "Action": [ "iam:CreatePolicyVersion", "iam:DeletePolicy", "iam:DeletePolicyVersion", "iam:SetDefaultPolicyVersion" ], "Resource": [ "arn:aws:iam::123456789012:policy/XCompanyBoundaries", "arn:aws:iam::123456789012:policy/DelegatedUserBoundary" ] }, { "Sid": "NoBoundaryUserDelete", "Effect": "Deny", "Action": "iam:DeleteUserPermissionsBoundary", "Resource": "*" } ] } ``` - **Blocks:** Lets a delegated administrator create or change IAM users only when the XCompanyBoundaries boundary is applied, and denies editing the boundary policies or deleting a user's boundary. - **Breaks:** AWS notes the delegated administrator can create a user who can access S3 resources the administrator cannot, so delegating effectively trusts the administrator with that access. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html)) - **Test before applying:** AWS documents that the IAM policy simulator can test IAM permissions boundaries, one boundary at a time. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) - **Mechanism:** [IAM permissions boundaries](/mechanisms/permissions-boundary/): A single IAM user or role it is attached to; it does not apply to other principals or to resources. - **Taken from:** [provider documentation](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html), provider documentation ## What the policy text names It names 46 actions or patterns: `iam:AttachUserPolicy`, `iam:CreateUser`, `iam:DeleteUserPolicy`, `iam:DetachUserPolicy`, `iam:PutUserPermissionsBoundary`, `iam:PutUserPolicy`, `cloudwatch:*`, `iam:CreateAccessKey`, and more. Its conditions test `iam:PermissionsBoundary`. ## What the source says "Sid: NoBoundaryUserDelete" (read 10 Nov 2026). On what it breaks: "By delegating these administrative actions, Maria effectively trusts Zhang with access to Amazon S3." ## Related guardrails - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege ## Related pages - [Allowed locations for new resources (Azure Azure Policy)](/guardrails/azure-policy-allowed-locations/) - [Allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-allowed-resource-types/) - [Not allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-not-allowed-resource-types/) ## Sources - Permissions boundary for delegated IAM user creation: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - How to test it first: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/aws-pb-delegated-user-creation-boundary/ --- # Allowed locations for new resources Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allowed locations for new resources is an Azure guardrail written in Azure Policy. Denies creation of resources in any location that is not on the allowed list you set in the policy parameter. What it breaks: deployments to other locations return 403, and the definition itself excludes resource groups, Microsoft.AzureActiveDirectory/b2cDirectories and resources that use the 'global' region. ```text Built-in policy definition: Allowed locations Definition ID: e56962a6-4747-49cd-b67b-bf8b01975c4c Category: General Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "location", "notIn": "[parameters('listOfAllowedLocations')]" }, { "field": "location", "notEquals": "global" }, { "field": "type", "notEquals": "Microsoft.AzureActiveDirectory/b2cDirectories" } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies creation of resources in any location that is not on the allowed list you set in the policy parameter. - **Breaks:** Deployments to other locations return 403, and the definition itself excludes resource groups, Microsoft.AzureActiveDirectory/b2cDirectories and resources that use the 'global' region. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedLocations_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedLocations_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy enables you to restrict the locations your organization can specify when deploying resources." (read 10 Nov 2026). On what it breaks: "Excludes resource groups, Microsoft.AzureActiveDirectory/b2cDirectories, and resources that use the 'global' region." ## Related guardrails - [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/): AWS, Service control policies (SCPs), region and residency - [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/): AWS, Service control policies (SCPs), region and residency - [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/): Google Cloud, Organization Policy Service (managed and legacy constraints), region and residency - [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/): Azure, Management group assignment, region and residency ## Related pages - [Allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-allowed-resource-types/) - [Not allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-not-allowed-resource-types/) - [Do not allow deletion of resource types (Azure Azure Policy)](/guardrails/azure-policy-deny-action-delete/) ## Sources - Allowed locations for new resources: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedLocations_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-allowed-locations/ --- # Allowed resource types Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allowed resource types is an Azure guardrail written in Azure Policy. Denies deployment of any resource type that is not on the allowed list you set in the policy parameter. What it breaks: only resource types that support tags and location are affected unless you duplicate the definition and change its mode to All. ```text Built-in policy definition: Allowed resource types Definition ID: a08ec900-254a-4555-9bf5-e42af04b5c5c Category: General Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "not": { "field": "type", "in": "[parameters('listOfResourceTypesAllowed')]" } }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies deployment of any resource type that is not on the allowed list you set in the policy parameter. - **Breaks:** Only resource types that support tags and location are affected unless you duplicate the definition and change its mode to All. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedResourceTypes_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedResourceTypes_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy enables you to specify the resource types that your organization can deploy." (read 10 Nov 2026). On what it breaks: "Only resource types that support 'tags' and 'location' will be affected by this policy." ## Related guardrails - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost - [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/): Azure, Azure Policy (built-in definitions), resource and cost ## Related pages - [Not allowed resource types (Azure Azure Policy)](/guardrails/azure-policy-not-allowed-resource-types/) - [Do not allow deletion of resource types (Azure Azure Policy)](/guardrails/azure-policy-deny-action-delete/) - [Allowed virtual machine size SKUs (Azure Azure Policy)](/guardrails/azure-policy-allowed-vm-sizes/) ## Sources - Allowed resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/AllowedResourceTypes_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-allowed-resource-types/ --- # Not allowed resource types Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Not allowed resource types is an Azure guardrail written in Azure Policy. Denies deployment of the resource types you list, for example expensive or high-risk services. What it breaks: a deny effect returns 403 for any create or update of a listed type, and compliance results are shown only for non-compliant resources. ```text Built-in policy definition: Not allowed resource types Definition ID: 6c112d4e-5bc7-47ae-a041-ea2d9dccd749 Category: General Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "in": "[parameters('listOfResourceTypesNotAllowed')]" }, { "value": "[field('type')]", "exists": true } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies deployment of the resource types you list, for example expensive or high-risk services. - **Breaks:** A deny effect returns 403 for any create or update of a listed type, and compliance results are shown only for non-compliant resources. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/InvalidResourceTypes_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/InvalidResourceTypes_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Restrict which resource types can be deployed in your environment." (read 10 Nov 2026). On what it breaks: "Compliance results are only shown for non-compliant resources." ## Related guardrails - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/): Azure, Azure Policy (built-in definitions), resource and cost ## Related pages - [Do not allow deletion of resource types (Azure Azure Policy)](/guardrails/azure-policy-deny-action-delete/) - [Allowed virtual machine size SKUs (Azure Azure Policy)](/guardrails/azure-policy-allowed-vm-sizes/) - [Network interfaces should not have public IPs (Azure Azure Policy)](/guardrails/azure-policy-nic-no-public-ip/) ## Sources - Not allowed resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/InvalidResourceTypes_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-not-allowed-resource-types/ --- # Do not allow deletion of resource types Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Do not allow deletion of resource types is an Azure guardrail written in Azure Policy. Blocks DELETE calls against the resource types you list, using the denyAction effect. What it breaks: the denyAction effect does not stop removal of child and extension resources when a delete targets the parent, and only supports the DELETE action. ```text Built-in policy definition: Do not allow deletion of resource types Definition ID: 78460a36-508a-49a4-b2b2-2f5ec564f4bb Category: General Supported effects: DenyAction, Disabled policyRule (copied from the repository file): { "if": { "field": "type", "in": "[parameters('listOfResourceTypesDisallowedForDeletion')]" }, "then": { "effect": "[parameters('effect')]", "details": { "actionNames": [ "delete" ] } } } ``` - **Blocks:** Blocks DELETE calls against the resource types you list, using the denyAction effect. - **Breaks:** The denyAction effect does not stop removal of child and extension resources when a delete targets the parent, and only supports the DELETE action. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny-action)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/ResourceType_denyActionDelete.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy enables you to specify the resource types that your organization can protect from accidentals deletion by blocking delete calls using deny action effect." (read 10 Nov 2026). On what it breaks: "Policy doesn't block removal of child and extension resources when a delete action targets the parent resources." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity ## Related pages - [Allowed virtual machine size SKUs (Azure Azure Policy)](/guardrails/azure-policy-allowed-vm-sizes/) - [Network interfaces should not have public IPs (Azure Azure Policy)](/guardrails/azure-policy-nic-no-public-ip/) - [Network interfaces should disable IP forwarding (Azure Azure Policy)](/guardrails/azure-policy-nic-no-ip-forwarding/) ## Sources - Do not allow deletion of resource types: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/General/ResourceType_denyActionDelete.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny-action (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-deny-action-delete/ --- # Allowed virtual machine size SKUs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allowed virtual machine size SKUs is an Azure guardrail written in Azure Policy. Denies creation of virtual machines whose size is not on the allowed list, which limits expensive VM sizes. What it breaks: requests for VM sizes outside the list are rejected with 403 at create or update time. ```text Built-in policy definition: Allowed virtual machine size SKUs Definition ID: cccc23c7-8427-4f53-ad12-b6a63eb452b3 Category: Compute Supported effects: Deny policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/virtualMachines" }, { "not": { "field": "Microsoft.Compute/virtualMachines/sku.name", "in": "[parameters('listOfAllowedSKUs')]" } } ] }, "then": { "effect": "Deny" } } ``` - **Blocks:** Denies creation of virtual machines whose size is not on the allowed list, which limits expensive VM sizes. - **Breaks:** Requests for VM sizes outside the list are rejected with 403 at create or update time. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/VMSkusAllowed_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy enables you to specify a set of virtual machine size SKUs that your organization can deploy." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost ## Related pages - [Network interfaces should not have public IPs (Azure Azure Policy)](/guardrails/azure-policy-nic-no-public-ip/) - [Network interfaces should disable IP forwarding (Azure Azure Policy)](/guardrails/azure-policy-nic-no-ip-forwarding/) - [Gateway subnets should not have a network security group (Azure Azure Policy)](/guardrails/azure-policy-gateway-subnet-no-nsg/) ## Sources - Allowed virtual machine size SKUs: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/VMSkusAllowed_Deny.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-allowed-vm-sizes/ --- # Network interfaces should not have public IPs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Network interfaces should not have public IPs is an Azure guardrail written in Azure Policy. Denies network interfaces that are configured with a public IP address. What it breaks: existing network interfaces that already have a public IP are marked non-compliant but not changed. ```text Built-in policy definition: Network interfaces should not have public IPs Definition ID: 83a86a26-fd1f-447c-b59d-e51f44264114 Category: Network Supported effects: deny policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Network/networkInterfaces" }, { "not": { "field": "Microsoft.Network/networkInterfaces/ipconfigurations[*].publicIpAddress.id", "notLike": "*" } } ] }, "then": { "effect": "deny" } } ``` - **Blocks:** Denies network interfaces that are configured with a public IP address. - **Breaks:** Existing network interfaces that already have a public IP are marked non-compliant but not changed. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkPublicIPNic_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy denies the network interfaces which are configured with any public IP." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure ## Related pages - [Network interfaces should disable IP forwarding (Azure Azure Policy)](/guardrails/azure-policy-nic-no-ip-forwarding/) - [Gateway subnets should not have a network security group (Azure Azure Policy)](/guardrails/azure-policy-gateway-subnet-no-nsg/) - [Configure storage accounts to disallow public blob access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-blob/) ## Sources - Network interfaces should not have public IPs: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkPublicIPNic_Deny.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-nic-no-public-ip/ --- # Network interfaces should disable IP forwarding Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Network interfaces should disable IP forwarding is an Azure guardrail written in Azure Policy. Denies network interfaces that have IP forwarding enabled, which turns off Azure's source and destination check. What it breaks: interfaces with IP forwarding enabled are rejected, and the definition says the setting should be reviewed by the network security team. ```text Built-in policy definition: Network interfaces should disable IP forwarding Definition ID: 88c0b9da-ce96-4b03-9635-f29a937e2900 Category: Network Supported effects: deny policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Network/networkInterfaces" }, { "field": "Microsoft.Network/networkInterfaces/enableIpForwarding", "equals": "true" } ] }, "then": { "effect": "deny" } } ``` - **Blocks:** Denies network interfaces that have IP forwarding enabled, which turns off Azure's source and destination check. - **Breaks:** Interfaces with IP forwarding enabled are rejected, and the definition says the setting should be reviewed by the network security team. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkIPForwardingNic_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkIPForwardingNic_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy denies the network interfaces which enabled IP forwarding." (read 10 Nov 2026). On what it breaks: "This should be reviewed by the network security team" ## Related guardrails - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter - [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/): Azure, Azure Policy (built-in definitions), network perimeter - [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/): Azure, Azure Policy (built-in definitions), network perimeter ## Related pages - [Gateway subnets should not have a network security group (Azure Azure Policy)](/guardrails/azure-policy-gateway-subnet-no-nsg/) - [Configure storage accounts to disallow public blob access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-blob/) - [Storage accounts should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-network/) ## Sources - Network interfaces should disable IP forwarding: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkIPForwardingNic_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-nic-no-ip-forwarding/ --- # Gateway subnets should not have a network security group Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Gateway subnets should not have a network security group is an Azure guardrail written in Azure Policy. Denies a network security group on a gateway subnet. What it breaks: you cannot filter gateway subnet traffic with an NSG while this applies; the definition states that an NSG on a gateway subnet makes the gateway stop functioning. ```text Built-in policy definition: Gateway subnets should not be configured with a network security group Definition ID: 35f9c03a-cc27-418e-9c0c-539ff999d010 Category: Network Supported effects: deny policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Network/virtualNetworks/subnets" }, { "field": "name", "equals": "GatewaySubnet" }, { "field": "Microsoft.Network/virtualNetworks/subnets/networkSecurityGroup.id", "exists": "true" } ] }, "then": { "effect": "deny" } } ``` - **Blocks:** Denies a network security group on a gateway subnet. - **Breaks:** You cannot filter gateway subnet traffic with an NSG while this applies; the definition states that an NSG on a gateway subnet makes the gateway stop functioning. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkSecurityGroupOnGatewaySubnet_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkSecurityGroupOnGatewaySubnet_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy denies if a gateway subnet is configured with a network security group." (read 10 Nov 2026). On what it breaks: "Assigning a network security group to a gateway subnet will cause the gateway to stop functioning." ## Related guardrails - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/): Azure, Azure Policy (built-in definitions), network perimeter ## Related pages - [Configure storage accounts to disallow public blob access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-blob/) - [Storage accounts should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-network/) - [Storage accounts should have the specified minimum TLS version (Azure Azure Policy)](/guardrails/azure-policy-storage-min-tls/) ## Sources - Gateway subnets should not have a network security group: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Network/NetworkSecurityGroupOnGatewaySubnet_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-gateway-subnet-no-nsg/ --- # Configure storage accounts to disallow public blob access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure storage accounts to disallow public blob access is an Azure guardrail written in Azure Policy. Sets the storage account public access property to disallowed on create or update, so containers and blobs cannot be opened to anonymous reads. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Configure your Storage account public access to be disallowed Definition ID: 13502221-8df0-4414-9937-de9c5c4e396b Category: Storage Supported effects: Modify, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Storage/storageAccounts" }, { "anyOf": [ { "allOf": [ { "value": "[requestContext().apiVersion]", "less": "2019-04-01" }, { "field": "Microsoft.Storage/storageAccounts/allowBlobPublicAccess", "exists": "true" } ] }, { "field": "Microsoft.Storage/storageAccounts/allowBlobPublicAccess", "equals": "true" } ] } ] }, "then": { "effect": "[parameters('effect')]", "details": { "conflictEffect": "audit", "roleDefinitionIds": [ "/providers/Microsoft.Authorization/roleDefinitions/17d1049b-9a84-46fb-8f53-869881c3d3ab" ], "operations": [ { "condition": "[greaterOrEquals(requestContext().apiVersion, '2019-04-01')]", "operation": "addOrReplace", "field": "Microsoft.Storage/storageAccounts/allowBlobPublicAccess", "value": false } ] } } } ``` - **Blocks:** Sets the storage account public access property to disallowed on create or update, so containers and blobs cannot be opened to anonymous reads. - **Breaks:** Modify rewrites the property during create or update instead of rejecting the request, and existing storage accounts change only through a remediation task. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountDisablePublicBlobAccess_Modify.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Anonymous public read access to containers and blobs in Azure Storage is a convenient way to share data but might present security risks." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure ## Related pages - [Storage accounts should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-storage-disable-public-network/) - [Storage accounts should have the specified minimum TLS version (Azure Azure Policy)](/guardrails/azure-policy-storage-min-tls/) - [Storage accounts should prevent shared key access (Azure Azure Policy)](/guardrails/azure-policy-storage-no-shared-key/) ## Sources - Configure storage accounts to disallow public blob access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountDisablePublicBlobAccess_Modify.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-storage-disable-public-blob/ --- # Storage accounts should disable public network access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Storage accounts should disable public network access is an Azure guardrail written in Azure Policy. Denies storage accounts that leave public network access enabled, so access is only through private endpoints. What it breaks: the documented effect of the setting is that it denies all logins that match IP or virtual network-based firewall rules. ```text Built-in policy definition: Storage accounts should disable public network access Definition ID: b2982f36-99f2-4db5-8eff-283140c09693 Category: Storage Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Storage/storageAccounts" }, { "not": { "field": "Microsoft.Storage/storageAccounts/publicNetworkAccess", "in": [ "Disabled", "SecuredByPerimeter" ] } } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies storage accounts that leave public network access enabled, so access is only through private endpoints. - **Breaks:** The documented effect of the setting is that it denies all logins that match IP or virtual network-based firewall rules. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StoragePublicNetworkAccess_AuditDeny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StoragePublicNetworkAccess_AuditDeny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "To improve the security of Storage Accounts, ensure that they aren't exposed to the public internet and can only be accessed from a private endpoint." (read 10 Nov 2026). On what it breaks: "This option disables access from any public address space outside the Azure IP range, and denies all logins that match IP or virtual network-based firewall rules." ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure ## Related pages - [Storage accounts should have the specified minimum TLS version (Azure Azure Policy)](/guardrails/azure-policy-storage-min-tls/) - [Storage accounts should prevent shared key access (Azure Azure Policy)](/guardrails/azure-policy-storage-no-shared-key/) - [Storage SAS tokens should adhere to 7 day maximum validity (Azure Azure Policy)](/guardrails/azure-policy-storage-sas-7-days/) ## Sources - Storage accounts should disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StoragePublicNetworkAccess_AuditDeny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-storage-disable-public-network/ --- # Storage accounts should have the specified minimum TLS version Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Storage accounts should have the specified minimum TLS version is an Azure guardrail written in Azure Policy. Denies storage accounts whose minimum TLS version is lower than the version you set (the definition recommends TLS 1.2 or newer). What it breaks: storage accounts below the minimum are rejected: deny returns 403 on create and update. ```text Built-in policy definition: Storage accounts should have the specified minimum TLS version Definition ID: fe83a0eb-a853-422d-aac2-1bffd182c5d0 Category: Storage Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Storage/storageAccounts" }, { "anyOf": [ { "field": "Microsoft.Storage/storageAccounts/minimumTlsVersion", "notEquals": "[parameters('minimumTlsVersion')]" }, { "field": "Microsoft.Storage/storageAccounts/minimumTlsVersion", "exists": "false" } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies storage accounts whose minimum TLS version is lower than the version you set (the definition recommends TLS 1.2 or newer). - **Breaks:** Storage accounts below the minimum are rejected: deny returns 403 on create and update. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountMinimumTLSVersion_Audit.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Configure a minimum TLS version for secure communication between the client application and the storage account." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys ## Related pages - [Storage accounts should prevent shared key access (Azure Azure Policy)](/guardrails/azure-policy-storage-no-shared-key/) - [Storage SAS tokens should adhere to 7 day maximum validity (Azure Azure Policy)](/guardrails/azure-policy-storage-sas-7-days/) - [Key vaults should have deletion protection enabled (Azure Azure Policy)](/guardrails/azure-policy-keyvault-purge-protection/) ## Sources - Storage accounts should have the specified minimum TLS version: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountMinimumTLSVersion_Audit.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-storage-min-tls/ --- # Storage accounts should prevent shared key access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Storage accounts should prevent shared key access is an Azure guardrail written in Azure Policy. Denies storage accounts that allow Shared Key authorization, so requests must use Microsoft Entra ID. What it breaks: existing storage accounts that allow Shared Key are marked non-compliant but not changed. ```text Built-in policy definition: Storage accounts should prevent shared key access Definition ID: 8c6a50c6-9ffd-4ae7-986f-5fa6111f9a54 Category: Storage Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Storage/storageAccounts" }, { "anyOf": [ { "value": "[contains(parameters('excludedManagedByResourceProviders'), 'Azure Red Hat OpenShift')]", "equals": false }, { "value": "[resourcegroup().managedBy]", "notContains": "/providers/Microsoft.RedHatOpenShift/" } ] }, { "anyOf": [ { "value": "[contains(parameters('excludedManagedByResourceProviders'), 'Azure Databricks')]", "equals": false }, { "value": "[resourcegroup().managedBy]", "notContains": "/providers/Microsoft.Databricks/" } ] }, { "anyOf": [ { "value": "[contains(parameters('excludedManagedByResourceProviders'), 'Microsoft Purview')]", "equals": false }, { "value": "[resourcegroup().managedBy]", "notContains": "/providers/Microsoft.Purview/" } ] }, { "anyOf": [ { "value": "[contains(parameters('excludedManagedByResourceProviders'), 'All managed resource groups')]", "equals": false }, { "value": "[empty(resourcegroup().managedBy)]", "equals": true } ] }, { "anyOf": [ { "anyOf": [ { "field": "Microsoft.Storage/storageAccounts/allowSharedKeyAccess", "exists": "false" }, { "field": "Microsoft.Storage/storageAccounts/allowSharedKeyAccess", "equals": "" } ] }, { "field": "Microsoft.Storage/storageAccounts/allowSharedKeyAccess", "equals": "true" } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies storage accounts that allow Shared Key authorization, so requests must use Microsoft Entra ID. - **Breaks:** Existing storage accounts that allow Shared Key are marked non-compliant but not changed. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountAllowSharedKeyAccess_Audit.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Audit requirement of Azure Active Directory (Azure AD) to authorize requests for your storage account." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege ## Related pages - [Storage SAS tokens should adhere to 7 day maximum validity (Azure Azure Policy)](/guardrails/azure-policy-storage-sas-7-days/) - [Key vaults should have deletion protection enabled (Azure Azure Policy)](/guardrails/azure-policy-keyvault-purge-protection/) - [Azure Key Vault should use the RBAC permission model (Azure Azure Policy)](/guardrails/azure-policy-keyvault-use-rbac/) ## Sources - Storage accounts should prevent shared key access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/StorageAccountAllowSharedKeyAccess_Audit.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-storage-no-shared-key/ --- # Storage SAS tokens should adhere to 7 day maximum validity Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Storage SAS tokens should adhere to 7 day maximum validity is an Azure guardrail written in Azure Policy. Denies storage accounts whose SAS expiration policy allows tokens valid for more than 7 days, which limits how long a leaked token can move data out. ```text Built-in policy definition: Storage SAS tokens should adhere to 7 day maximum validity Definition ID: 7aa1c9d5-3d7e-4579-8117-d85e99211757 Category: Storage Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Storage/storageAccounts" }, { "anyOf": [ { "field": "Microsoft.Storage/storageAccounts/sasPolicy", "exists": "false" }, { "field": "Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod", "exists": "false" }, { "value": "[if(empty(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod')), 99, int(split(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod'), createArray(':', '.'))[0]))]", "greater": 7 }, { "allOf": [ { "value": "[if(empty(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod')), 0, int(split(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod'), createArray(':', '.'))[0]))]", "equals": 7 }, { "anyOf": [ { "value": "[if(empty(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod')), 0, int(split(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod'), createArray(':', '.'))[1]))]", "notEquals": 0 }, { "value": "[if(empty(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod')), 0, int(split(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod'), createArray(':', '.'))[2]))]", "notEquals": 0 }, { "value": "[if(empty(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod')), 0, int(split(field('Microsoft.Storage/storageAccounts/sasPolicy.sasExpirationPeriod'), createArray(':', '.'))[3]))]", "notEquals": 0 } ] } ] }, { "field": "Microsoft.Storage/storageAccounts/sasPolicy.expirationAction", "exists": "false" }, { "field": "Microsoft.Storage/storageAccounts/sasPolicy.expirationAction", "notEquals": "Block" } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies storage accounts whose SAS expiration policy allows tokens valid for more than 7 days, which limits how long a leaked token can move data out. - **Breaks:** Existing storage accounts with longer SAS lifetimes are marked non-compliant but not changed. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/EnforceStorageSasMaxExpiration7Days_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "This policy ensures that Shared Access Signature (SAS) tokens for storage accounts are configured with a maximum validity period of 7 days or less." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration ## Related pages - [Key vaults should have deletion protection enabled (Azure Azure Policy)](/guardrails/azure-policy-keyvault-purge-protection/) - [Azure Key Vault should use the RBAC permission model (Azure Azure Policy)](/guardrails/azure-policy-keyvault-use-rbac/) - [Configure key vaults to enable the firewall (Azure Azure Policy)](/guardrails/azure-policy-keyvault-firewall/) ## Sources - Storage SAS tokens should adhere to 7 day maximum validity: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Storage/EnforceStorageSasMaxExpiration7Days_Deny.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-storage-sas-7-days/ --- # Key vaults should have deletion protection enabled Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Key vaults should have deletion protection enabled is an Azure guardrail written in Azure Policy. Denies key vaults without purge protection and soft delete enabled, so keys and secrets cannot be purged during the retention period. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Key vaults should have deletion protection enabled Definition ID: 0b60c0b2-2dc2-4e1c-b5c9-abbed971de53 Category: Key Vault Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.KeyVault/vaults" }, { "field": "Microsoft.KeyVault/vaults/createMode", "notEquals": "recover" }, { "anyOf": [ { "allOf": [ { "value": "[requestContext().apiVersion]", "less": "2019-09-01" }, { "anyOf": [ { "field": "Microsoft.KeyVault/vaults/enableSoftDelete", "equals": "false" }, { "field": "Microsoft.KeyVault/vaults/enableSoftDelete", "exists": "false" } ] } ] }, { "allOf": [ { "value": "[requestContext().apiVersion]", "greaterOrEquals": "2019-09-01" }, { "field": "Microsoft.KeyVault/vaults/enableSoftDelete", "equals": "false" } ] }, { "field": "Microsoft.KeyVault/vaults/enablePurgeProtection", "exists": "false" }, { "field": "Microsoft.KeyVault/vaults/enablePurgeProtection", "equals": "false" } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies key vaults without purge protection and soft delete enabled, so keys and secrets cannot be purged during the retention period. - **Breaks:** Purge protection enforces a mandatory retention period: no one in your organization or Microsoft can purge the vault during the soft delete retention period. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/Recoverable_Audit.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/Recoverable_Audit.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Malicious deletion of a key vault can lead to permanent data loss." (read 10 Nov 2026). On what it breaks: "No one inside your organization or Microsoft will be able to purge your key vaults during the soft delete retention period." ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys ## Related pages - [Azure Key Vault should use the RBAC permission model (Azure Azure Policy)](/guardrails/azure-policy-keyvault-use-rbac/) - [Configure key vaults to enable the firewall (Azure Azure Policy)](/guardrails/azure-policy-keyvault-firewall/) - [OS and data disks should be encrypted with a customer-managed key (Azure Azure Policy)](/guardrails/azure-policy-disk-cmk/) ## Sources - Key vaults should have deletion protection enabled: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/Recoverable_Audit.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-keyvault-purge-protection/ --- # Azure Key Vault should use the RBAC permission model Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure Key Vault should use the RBAC permission model is an Azure guardrail written in Azure Policy. Denies key vaults that use the legacy access policy model instead of Azure RBAC. What it breaks: existing key vaults on access policies are marked non-compliant but not changed. ```text Built-in policy definition: Azure Key Vault should use RBAC permission model Definition ID: 12d4fa5e-1f9f-4c21-97a9-b99b3c6611b5 Category: Key Vault Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.KeyVault/vaults" }, { "not": { "field": "Microsoft.KeyVault/vaults/createMode", "equals": "recover" } }, { "anyOf": [ { "field": "Microsoft.KeyVault/vaults/enableRbacAuthorization", "exists": "false" }, { "field": "Microsoft.KeyVault/vaults/enableRbacAuthorization", "equals": "false" } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies key vaults that use the legacy access policy model instead of Azure RBAC. - **Breaks:** Existing key vaults on access policies are marked non-compliant but not changed. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/Should_Use_RBAC.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Enable RBAC permission model across Key Vaults." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege ## Related pages - [Configure key vaults to enable the firewall (Azure Azure Policy)](/guardrails/azure-policy-keyvault-firewall/) - [OS and data disks should be encrypted with a customer-managed key (Azure Azure Policy)](/guardrails/azure-policy-disk-cmk/) - [Azure SQL Database should have Microsoft Entra-only authentication enabled (Azure Azure Policy)](/guardrails/azure-policy-sql-entra-only/) ## Sources - Azure Key Vault should use the RBAC permission model: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/Should_Use_RBAC.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-keyvault-use-rbac/ --- # Configure key vaults to enable the firewall Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure key vaults to enable the firewall is an Azure guardrail written in Azure Policy. Turns on the key vault firewall at create or update so the vault is not reachable by default from public IPs. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Configure key vaults to enable firewall Definition ID: ac673a9a-f77d-4846-b2d8-a57f8e1c01dc Category: Key Vault Supported effects: Modify, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.KeyVault/vaults" }, { "field": "Microsoft.KeyVault/vaults/networkAcls.defaultAction", "notEquals": "Deny" } ] }, "then": { "effect": "[parameters('effect')]", "details": { "conflictEffect": "audit", "roleDefinitionIds": [ "/providers/Microsoft.Authorization/roleDefinitions/f25e0fa2-a7c8-4377-a976-54943a77a395" ], "operations": [ { "operation": "addOrReplace", "field": "Microsoft.KeyVault/vaults/networkAcls.defaultAction", "value": "Deny" } ] } } } ``` - **Blocks:** Turns on the key vault firewall at create or update so the vault is not reachable by default from public IPs. - **Breaks:** Modify changes the property during create or update instead of rejecting the request, and existing vaults change only through a remediation task. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/FirewallEnabled_Modify.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Enable the key vault firewall so that the key vault is not accessible by default to any public IPs." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/): Azure, Azure Policy (built-in definitions), network perimeter ## Related pages - [OS and data disks should be encrypted with a customer-managed key (Azure Azure Policy)](/guardrails/azure-policy-disk-cmk/) - [Azure SQL Database should have Microsoft Entra-only authentication enabled (Azure Azure Policy)](/guardrails/azure-policy-sql-entra-only/) - [Deploy diagnostic settings for Key Vault to Event Hub (Azure Azure Policy)](/guardrails/azure-policy-keyvault-diagnostic-logs/) ## Sources - Configure key vaults to enable the firewall: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/FirewallEnabled_Modify.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-keyvault-firewall/ --- # OS and data disks should be encrypted with a customer-managed key Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. OS and data disks should be encrypted with a customer-managed key is an Azure guardrail written in Azure Policy. Denies managed disks that are not encrypted with a customer-managed key. What it breaks: you take on key lifecycle work: the definition says you have full control and responsibility for the key lifecycle, including rotation and management. ```text Built-in policy definition: OS and data disks should be encrypted with a customer-managed key Definition ID: 702dd420-7fcc-42c5-afe8-4026edd20fe0 Category: Compute Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "anyOf": [ { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/virtualMachines" }, { "field": "Microsoft.Compute/virtualMachines/storageProfile.osDisk.managedDisk.diskEncryptionSet.id", "exists": "False" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/virtualMachines" }, { "value": "[length(field('Microsoft.Compute/virtualMachines/storageProfile.dataDisks'))]", "greater": 0 }, { "field": "Microsoft.Compute/virtualMachines/storageProfile.dataDisks[*].managedDisk.id", "exists": "False" }, { "field": "Microsoft.Compute/virtualMachines/storageProfile.dataDisks[*].managedDisk.diskEncryptionSet.id", "exists": "False" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/virtualMachineScaleSets" }, { "field": "Microsoft.Compute/virtualMachineScaleSets/virtualMachineProfile.storageProfile.osDisk.managedDisk.diskEncryptionSet.id", "exists": "False" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/virtualMachineScaleSets" }, { "count": { "field": "Microsoft.Compute/virtualMachineScaleSets/virtualMachineProfile.storageProfile.dataDisks[*]" }, "greater": 0 }, { "not": { "field": "Microsoft.Compute/virtualMachineScaleSets/virtualMachineProfile.storageProfile.dataDisks[*].managedDisk.diskEncryptionSet.id", "exists": "true" } } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/disks" }, { "field": "Microsoft.Compute/disks/encryption.diskEncryptionSetId", "exists": "False" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/galleries/images/versions" }, { "value": "[length(field('Microsoft.Compute/galleries/images/versions/publishingProfile.targetRegions[*].encryption.osDiskImage.diskEncryptionSetId'))]", "notEquals": "[length(field('Microsoft.Compute/galleries/images/versions/publishingProfile.targetRegions[*]'))]" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/galleries/images/versions" }, { "value": "[length(field('Microsoft.Compute/galleries/images/versions/storageProfile.dataDiskImages[*]'))]", "greater": 0 }, { "anyOf": [ { "count": { "field": "Microsoft.Compute/galleries/images/versions/publishingProfile.targetRegions[*]", "where": { "value": "[length(current('Microsoft.Compute/galleries/images/versions/publishingProfile.targetRegions[*].encryption.dataDiskImages[*].diskEncryptionSetId'))]", "notEquals": "[length(field('Microsoft.Compute/galleries/images/versions/storageProfile.dataDiskImages[*]'))]" } }, "greater": 0 }, { "not": { "field": "Microsoft.Compute/galleries/images/versions/publishingProfile.targetRegions[*].encryption.dataDiskImages[*].diskEncryptionSetId", "exists": "true" } } ] } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/images" }, { "field": "Microsoft.Compute/images/storageProfile.osDisk.diskEncryptionSet.id", "exists": "False" } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.Compute/images" }, { "value": "[length(field('Microsoft.Compute/images/storageProfile.dataDisks[*]'))]", "greater": 0 }, { "not": { "field": "Microsoft.Compute/images/storageProfile.dataDisks[*].diskEncryptionSet.id", "exists": "true" } } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies managed disks that are not encrypted with a customer-managed key. - **Breaks:** You take on key lifecycle work: the definition says you have full control and responsibility for the key lifecycle, including rotation and management. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/OSAndDataDiskCMKRequired_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/OSAndDataDiskCMKRequired_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Use customer-managed keys to manage the encryption at rest of the contents of your managed disks." (read 10 Nov 2026). On what it breaks: "You have full control and responsibility for the key lifecycle, including rotation and management." ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys ## Related pages - [Azure SQL Database should have Microsoft Entra-only authentication enabled (Azure Azure Policy)](/guardrails/azure-policy-sql-entra-only/) - [Deploy diagnostic settings for Key Vault to Event Hub (Azure Azure Policy)](/guardrails/azure-policy-keyvault-diagnostic-logs/) - [Application Insights components should block non-Entra ingestion (Azure Azure Policy)](/guardrails/azure-policy-appinsights-entra-ingestion/) ## Sources - OS and data disks should be encrypted with a customer-managed key: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Compute/OSAndDataDiskCMKRequired_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-disk-cmk/ --- # Azure SQL Database should have Microsoft Entra-only authentication enabled Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure SQL Database should have Microsoft Entra-only authentication enabled is an Azure guardrail written in Azure Policy. Denies enabling local (SQL) authentication on Azure SQL logical servers after they are created. What it breaks: the definition does not block servers from being created with local authentication enabled; Microsoft points to the Entra-only authentication initiative to cover both. ```text Built-in policy definition: Azure SQL Database should have Microsoft Entra-only authentication enabled Definition ID: b3a22bc9-66de-45fb-98fa-00f5df42f41a Category: SQL Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Sql/servers/azureADOnlyAuthentications" }, { "value": "[resourcegroup().managedBy]", "notContains": "/providers/Microsoft.Synapse/" }, { "anyOf": [ { "field": "Microsoft.Sql/servers/azureADOnlyAuthentications/azureADOnlyAuthentication", "exists": false }, { "field": "Microsoft.Sql/servers/azureADOnlyAuthentications/azureADOnlyAuthentication", "equals": false } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies enabling local (SQL) authentication on Azure SQL logical servers after they are created. - **Breaks:** The definition does not block servers from being created with local authentication enabled; Microsoft points to the Entra-only authentication initiative to cover both. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/SQL/SqlServer_ADOnlyEnabled_DisableADAuth_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/SQL/SqlServer_ADOnlyEnabled_DisableADAuth_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Require Azure SQL logical servers to use Microsoft Entra-only authentication." (read 10 Nov 2026). On what it breaks: "This policy doesn't block servers from being created with local authentication enabled." ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege ## Related pages - [Deploy diagnostic settings for Key Vault to Event Hub (Azure Azure Policy)](/guardrails/azure-policy-keyvault-diagnostic-logs/) - [Application Insights components should block non-Entra ingestion (Azure Azure Policy)](/guardrails/azure-policy-appinsights-entra-ingestion/) - [Configure Cognitive Services accounts to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-public-network/) ## Sources - Azure SQL Database should have Microsoft Entra-only authentication enabled: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/SQL/SqlServer_ADOnlyEnabled_DisableADAuth_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-sql-entra-only/ --- # Deploy diagnostic settings for Key Vault to Event Hub Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deploy diagnostic settings for Key Vault to Event Hub is an Azure guardrail written in Azure Policy. Deploys diagnostic settings that stream key vault logs to an Event Hub whenever a vault without them is created or updated. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Deploy Diagnostic Settings for Key Vault to Event Hub Definition ID: ed7c8c13-51e7-49d1-8a43-8490431a0da2 Category: Key Vault Supported effects: DeployIfNotExists, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.KeyVault/vaults" }, { "anyOf": [ { "value": "[parameters('eventHubLocation')]", "equals": "" }, { "field": "location", "equals": "[parameters('eventHubLocation')]" } ] } ] }, "then": { "effect": "[parameters('effect')]", "details": { "type": "Microsoft.Insights/diagnosticSettings", "name": "[parameters('profileName')]", "existenceCondition": { "allOf": [ { "count": { "field": "Microsoft.Insights/diagnosticSettings/logs[*]", "where": { "field": "Microsoft.Insights/diagnosticSettings/logs[*].enabled", "equals": "[parameters('logsEnabled')]" } }, "greaterOrEquals": 1 }, { "count": { "field": "Microsoft.Insights/diagnosticSettings/metrics[*]", "where": { "field": "Microsoft.Insights/diagnosticSettings/metrics[*].enabled", "equals": "[parameters('metricsEnabled')]" } }, "greaterOrEquals": 1 } ] }, "roleDefinitionIds": [ "/providers/microsoft.authorization/roleDefinitions/b24988ac-6180-42a0-ab88-20f7382dd24c" ], "deployment": { "properties": { "mode": "incremental", "template": { "$schema": "http://schema.management.azure.com/schemas/2015-01-01/deploymentTemplate.json#", "contentVersion": "1.0.0.0", "parameters": { "vaultName": { "type": "string" }, "location": { "type": "string" }, "eventHubRuleId": { "type": "string" }, "metricsEnabled": { "type": "string" }, "logsEnabled": { "type": "string" }, "profileName": { "type": "string" } }, "resources": [ { "type": "Microsoft.KeyVault/vaults/providers/diagnosticSettings", "apiVersion": "2017-05-01-preview", "name": "[concat(parameters('vaultName'), '/', 'Microsoft.Insights/', parameters('profileName'))]", "location": "[parameters('location')]", "dependsOn": [], "properties": { "eventHubAuthorizationRuleId": "[parameters('eventHubRuleId')]", "metrics": [ { "category": "AllMetrics", "enabled": "[parameters('metricsEnabled')]", "retentionPolicy": { "enabled": false, "days": 0 } } ], "logs": [ { "category": "AuditEvent", "enabled": "[parameters('logsEnabled')]" }, { "category": "AzurePolicyEvaluationDetails", "enabled": "[parameters('logsEnabled')]" } ] } } ], "outputs": { "policy": { "type": "string", "value": "[concat('Enabled diagnostic settings for ', parameters('vaultName'))]" } } }, "parameters": { "location": { "value": "[field('location')]" }, "vaultName": { "value": "[field('name')]" }, "eventHubRuleId": { "value": "[parameters('eventHubRuleId')]" }, "metricsEnabled": { "value": "[parameters('metricsEnabled')]" }, "logsEnabled": { "value": "[parameters('logsEnabled')]" }, "profileName": { "value": "[parameters('profileName')]" } } } } } } } ``` - **Blocks:** Deploys diagnostic settings that stream key vault logs to an Event Hub whenever a vault without them is created or updated. - **Breaks:** DeployIfNotExists is not a blocking effect: matching resources are marked non-compliant and no action is taken on them until a remediation task runs. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deploy-if-not-exists)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/DiagnosticLog_DINE.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Deploys the diagnostic settings for Key Vault to stream to a regional Event Hub when any Key Vault which is missing this diagnostic settings is created or updated." (read 10 Nov 2026). On what it breaks: "During an evaluation cycle, policy definitions with a DeployIfNotExists effect that match resources are marked as non-compliant, but no action is taken on that resource." ## Related guardrails - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring ## Related pages - [Application Insights components should block non-Entra ingestion (Azure Azure Policy)](/guardrails/azure-policy-appinsights-entra-ingestion/) - [Configure Cognitive Services accounts to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-public-network/) - [Configure Cognitive Services accounts to disable local authentication (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth/) ## Sources - Deploy diagnostic settings for Key Vault to Event Hub: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Key%20Vault/DiagnosticLog_DINE.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deploy-if-not-exists (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-keyvault-diagnostic-logs/ --- # Application Insights components should block non-Entra ingestion Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Application Insights components should block non-Entra ingestion is an Azure guardrail written in Azure Policy. Denies Application Insights components that accept log ingestion without Microsoft Entra authentication. What it breaks: existing components that accept unauthenticated ingestion are marked non-compliant but not changed. ```text Built-in policy definition: Application Insights components should block non-Azure Active Directory based ingestion. Definition ID: 199d5677-e4d9-4264-9465-efe1839c06bd Category: Monitoring Supported effects: Deny, Audit, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Insights/components" }, { "field": "Microsoft.Insights/components/DisableLocalAuth", "notEquals": "true" } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies Application Insights components that accept log ingestion without Microsoft Entra authentication. - **Breaks:** Existing components that accept unauthenticated ingestion are marked non-compliant but not changed. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Monitoring/ApplicationInsightsComponents_DisableLocalAuth_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Enforcing log ingestion to require Azure Active Directory authentication prevents unauthenticated logs from an attacker which could lead to incorrect status, false alerts, and incorrect logs stored in the system." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring ## Related pages - [Configure Cognitive Services accounts to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-public-network/) - [Configure Cognitive Services accounts to disable local authentication (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth/) - [Configure Azure AI Services resources to disable local key access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth-dine/) ## Sources - Application Insights components should block non-Entra ingestion: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Monitoring/ApplicationInsightsComponents_DisableLocalAuth_Deny.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-appinsights-entra-ingestion/ --- # Configure Cognitive Services accounts to disable public network access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure Cognitive Services accounts to disable public network access is an Azure guardrail written in Azure Policy. Sets public network access to disabled on Cognitive Services (Azure AI services, including Azure OpenAI) accounts on create or update. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Configure Cognitive Services accounts to disable public network access Definition ID: 47ba1dd7-28d9-4b07-a8d5-9813bed64e0c Category: Cognitive Services Supported effects: Disabled, Modify policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.CognitiveServices/accounts" }, { "field": "Microsoft.CognitiveServices/accounts/publicNetworkAccess", "notEquals": "Disabled" } ] }, "then": { "effect": "[parameters('effect')]", "details": { "roleDefinitionIds": [ "/providers/microsoft.authorization/roleDefinitions/b24988ac-6180-42a0-ab88-20f7382dd24c" ], "conflictEffect": "audit", "operations": [ { "condition": "[greaterOrEquals(requestContext().apiVersion, '2017-04-18')]", "operation": "addOrReplace", "field": "Microsoft.CognitiveServices/accounts/publicNetworkAccess", "value": "Disabled" } ] } } } ``` - **Blocks:** Sets public network access to disabled on Cognitive Services (Azure AI services, including Azure OpenAI) accounts on create or update. - **Breaks:** Modify changes the property during create or update instead of rejecting the request, and existing accounts change only through a remediation task. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DisablePublicNetworkAccess_Modify.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Disable public network access for your Cognitive Services resource so that it's not accessible over the public internet." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Configure Cognitive Services accounts to disable local authentication (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth/) - [Configure Azure AI Services resources to disable local key access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth-dine/) - [Configure Azure Machine Learning workspaces to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ml-workspace-disable-public-network/) ## Sources - Configure Cognitive Services accounts to disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DisablePublicNetworkAccess_Modify.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-ai-disable-public-network/ --- # Configure Cognitive Services accounts to disable local authentication Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure Cognitive Services accounts to disable local authentication is an Azure guardrail written in Azure Policy. Disables key-based (local) authentication on Cognitive Services accounts so only Microsoft Entra identities can call them. What it breaks: modify changes the property during create or update instead of rejecting the request, and existing accounts change only through a remediation task. ```text Built-in policy definition: Configure Cognitive Services accounts to disable local authentication methods Definition ID: 14de9e63-1b31-492e-a5a3-c3f7fd57f555 Category: Cognitive Services Supported effects: Modify, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.CognitiveServices/accounts" }, { "field": "Microsoft.CognitiveServices/accounts/disableLocalAuth", "notEquals": true } ] }, "then": { "effect": "[parameters('effect')]", "details": { "roleDefinitionIds": [ "/providers/microsoft.authorization/roleDefinitions/b24988ac-6180-42a0-ab88-20f7382dd24c" ], "conflictEffect": "audit", "operations": [ { "condition": "[greaterOrEquals(requestContext().apiVersion, '2021-04-30')]", "operation": "addOrReplace", "field": "Microsoft.CognitiveServices/accounts/disableLocalAuth", "value": true } ] } } } ``` - **Blocks:** Disables key-based (local) authentication on Cognitive Services accounts so only Microsoft Entra identities can call them. - **Breaks:** Modify changes the property during create or update instead of rejecting the request, and existing accounts change only through a remediation task. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DisableLocalAuth_Modify.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Disable local authentication methods so that your Cognitive Services accounts require Azure Active Directory identities exclusively for authentication." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Configure Azure AI Services resources to disable local key access (Azure Azure Policy)](/guardrails/azure-policy-ai-disable-local-auth-dine/) - [Configure Azure Machine Learning workspaces to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ml-workspace-disable-public-network/) - [Azure Machine Learning registry deployments restricted to the allowed registry (preview) (Azure Azure Policy)](/guardrails/azure-policy-ml-registry-models/) ## Sources - Configure Cognitive Services accounts to disable local authentication: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DisableLocalAuth_Modify.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-ai-disable-local-auth/ --- # Configure Azure AI Services resources to disable local key access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure Azure AI Services resources to disable local key access is an Azure guardrail written in Azure Policy. Deploys the setting that disables key access on Azure AI Services resources, leaving Microsoft Entra ID as the only access method. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Configure Azure AI Services resources to disable local key access (disable local authentication) Definition ID: 55eff01b-f2bd-4c32-9203-db285f709d30 Category: Azure Ai Services Supported effects: DeployIfNotExists, Disabled policyRule (copied from the repository file): { "if": { "field": "type", "equals": "Microsoft.CognitiveServices/accounts" }, "then": { "effect": "[parameters('effect')]", "details": { "type": "Microsoft.CognitiveServices/accounts", "name": "[field('name')]", "existenceCondition": { "field": "Microsoft.CognitiveServices/accounts/disableLocalAuth", "equals": true }, "roleDefinitionIds": [ "/providers/Microsoft.Authorization/roleDefinitions/a001fd3d-188f-4b5d-821b-7da978bf7442", "/providers/microsoft.authorization/roleDefinitions/25fbc0a9-bd7c-42a3-aa1a-3b75d497ee68" ], "deployment": { "properties": { "mode": "incremental", "template": { "$schema": "http://schema.management.azure.com/schemas/2015-01-01/deploymentTemplate.json#", "contentVersion": "1.0.0.0", "parameters": { "resourceName": { "type": "string" }, "location": { "type": "string" }, "disableLocalAuth": { "type": "string" }, "kind": { "type": "string" }, "sku": { "type": "object" }, "publicNetworkAccess": { "type": "string" }, "networkAcls": { "type": "object" } }, "variables": {}, "resources": [ { "type": "Microsoft.CognitiveServices/accounts", "apiVersion": "2023-10-01-preview", "name": "[parameters('resourceName')]", "kind": "[parameters('kind')]", "sku": "[parameters('sku')]", "location": "[parameters('location')]", "properties": { "customSubDomainName": "[parameters('resourceName')]", "disableLocalAuth": "[parameters('disableLocalAuth')]", "publicNetworkAccess": "[parameters('publicNetworkAccess')]", "networkAcls": "[parameters('networkAcls')]" } } ], "outputs": {} }, "parameters": { "location": { "value": "[field('location')]" }, "resourceName": { "value": "[field('name')]" }, "disableLocalAuth": { "value": "[parameters('disableLocalAuth')]" }, "kind": { "value": "[field('kind')]" }, "sku": { "value": "[field('Microsoft.CognitiveServices/accounts/sku')]" }, "publicNetworkAccess": { "value": "[parameters('publicNetworkAccess')]" }, "networkAcls": { "value": "[field('Microsoft.CognitiveServices/accounts/networkAcls')]" } } } } } } } ``` - **Blocks:** Deploys the setting that disables key access on Azure AI Services resources, leaving Microsoft Entra ID as the only access method. - **Breaks:** Azure OpenAI Studio requires key access and will not function if key access is disabled, per the definition description. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Azure%20Ai%20Services/DisableLocalAuth_CognitiveServices_DINE.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Azure%20Ai%20Services/DisableLocalAuth_CognitiveServices_DINE.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Key access (local authentication) is recommended to be disabled for security." (read 10 Nov 2026). On what it breaks: "Azure OpenAI Studio, typically used in development/testing, requires key access and will not function if key access is disabled." ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Configure Azure Machine Learning workspaces to disable public network access (Azure Azure Policy)](/guardrails/azure-policy-ml-workspace-disable-public-network/) - [Azure Machine Learning registry deployments restricted to the allowed registry (preview) (Azure Azure Policy)](/guardrails/azure-policy-ml-registry-models/) - [Microsoft Foundry agents and toolboxes should not use denied tools (preview) (Azure Azure Policy)](/guardrails/azure-policy-foundry-denied-tools/) ## Sources - Configure Azure AI Services resources to disable local key access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Azure%20Ai%20Services/DisableLocalAuth_CognitiveServices_DINE.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-ai-disable-local-auth-dine/ --- # Configure Azure Machine Learning workspaces to disable public network access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Configure Azure Machine Learning workspaces to disable public network access is an Azure guardrail written in Azure Policy. Sets public network access to disabled on Azure Machine Learning workspaces on create or update. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: Configure Azure Machine Learning Workspaces to disable public network access Definition ID: a10ee784-7409-4941-b091-663697637c0f Category: Machine Learning Supported effects: Modify, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.MachineLearningServices/workspaces" }, { "field": "Microsoft.MachineLearningServices/workspaces/publicNetworkAccess", "notEquals": "Disabled" } ] }, "then": { "effect": "[parameters('effect')]", "details": { "roleDefinitionIds": [ "/providers/microsoft.authorization/roleDefinitions/b24988ac-6180-42a0-ab88-20f7382dd24c" ], "conflictEffect": "audit", "operations": [ { "condition": "[greaterOrEquals(requestContext().apiVersion, '2021-10-01')]", "operation": "addOrReplace", "field": "Microsoft.MachineLearningServices/workspaces/publicNetworkAccess", "value": "Disabled" } ] } } } ``` - **Blocks:** Sets public network access to disabled on Azure Machine Learning workspaces on create or update. - **Breaks:** Modify changes the property during create or update instead of rejecting the request, and existing workspaces change only through a remediation task; Microsoft suggests private endpoints to control exposure instead. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Machine%20Learning/Workspace_PublicNetworkAccessDisabled_Modify.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Disable public network access for Azure Machine Learning Workspaces so that your workspaces aren't accessible over the public internet." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Azure Machine Learning registry deployments restricted to the allowed registry (preview) (Azure Azure Policy)](/guardrails/azure-policy-ml-registry-models/) - [Microsoft Foundry agents and toolboxes should not use denied tools (preview) (Azure Azure Policy)](/guardrails/azure-policy-foundry-denied-tools/) - [Azure AI Search services should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-search-disable-public-network/) ## Sources - Configure Azure Machine Learning workspaces to disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Machine%20Learning/Workspace_PublicNetworkAccessDisabled_Modify.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-modify (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-ml-workspace-disable-public-network/ --- # Azure Machine Learning registry deployments restricted to the allowed registry (preview) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure Machine Learning registry deployments restricted to the allowed registry (preview) is an Azure guardrail written in Azure Policy. Denies deployments of registry models unless they come from the allowed registry and are not on the restricted model list. Assign the definition with enforcementMode set to DoNotEnforce, which Microsoft documents as the way to evaluate an assignment before enforcement. ```text Built-in policy definition: [Preview]: Azure Machine Learning Model Registry Deployments are restricted except for the allowed Registry Definition ID: 19539b54-c61e-4196-9a38-67598701be90 Category: Machine Learning Supported effects: Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.MachineLearningServices.v2.Data/workspaces/deployments" }, { "anyOf": [ { "field": "Microsoft.MachineLearningServices.v2.Data/workspaces/deployments/model.assetId", "in": "[parameters('restrictedModels')]" }, { "allOf": [ { "field": "Microsoft.MachineLearningServices.v2.Data/workspaces/deployments/model.assetId", "like": "azureml://registries/*" }, { "not": { "field": "Microsoft.MachineLearningServices.v2.Data/workspaces/deployments/model.assetId", "like": "[concat('azureml://registries/', parameters('allowedRegistryName'), '/*')]" } } ] } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies deployments of registry models unless they come from the allowed registry and are not on the restricted model list. - **Breaks:** This definition is a preview, and if no allowed registry name is set the parameter description says all models from all registries are restricted. ([source](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Machine%20Learning/OnlyDeployAllowedRegistryModels_Deny.json)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Machine%20Learning/OnlyDeployAllowedRegistryModels_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Only deploy Registry Models in the allowed Registry and that are not restricted." (read 10 Nov 2026). On what it breaks: "If not set, by default all models from all registries will be restricted." ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Microsoft Foundry agents and toolboxes should not use denied tools (preview) (Azure Azure Policy)](/guardrails/azure-policy-foundry-denied-tools/) - [Azure AI Search services should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-search-disable-public-network/) - [Deployment stack deny settings (Azure RBAC deny assignment) (Azure Azure RBAC deny assignments)](/guardrails/azure-deny-assignment-deployment-stack/) ## Sources - Azure Machine Learning registry deployments restricted to the allowed registry (preview): https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Machine%20Learning/OnlyDeployAllowedRegistryModels_Deny.json (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-ml-registry-models/ --- # Microsoft Foundry agents and toolboxes should not use denied tools (preview) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Microsoft Foundry agents and toolboxes should not use denied tools (preview) is an Azure guardrail written in Azure Policy. Denies Microsoft Foundry agents and toolboxes that use tool types, MCP endpoints or MCP connectors on your denied lists. Copied from a licensed open source set and read on 10 Nov 2026. ```text Built-in policy definition: [Preview]: Microsoft Foundry agents and toolboxes should not use denied tools Definition ID: 7740c22b-07ef-4bac-9f5a-e38d188f80b7 Category: Cognitive Services Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "anyOf": [ { "allOf": [ { "field": "type", "equals": "Microsoft.CognitiveServices.Data/accounts/projects/agents" }, { "count": { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*]", "where": { "anyOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*].type", "in": "[parameters('deniedToolTypes')]" }, { "allOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*].type", "equals": "mcp" }, { "count": { "value": "[parameters('deniedMcpEndpoints')]", "name": "endpointPatternAgents", "where": { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*].serverUrl", "like": "[current('endpointPatternAgents')]" } }, "greater": 0 } ] }, { "allOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*].type", "equals": "mcp" }, { "field": "Microsoft.CognitiveServices.Data/accounts/projects/agents/tools[*].connectorId", "in": "[parameters('deniedMcpConnectors')]" } ] } ] } }, "greater": 0 } ] }, { "allOf": [ { "field": "type", "equals": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes" }, { "count": { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*]", "where": { "anyOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*].type", "in": "[parameters('deniedToolTypes')]" }, { "allOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*].type", "equals": "mcp" }, { "count": { "value": "[parameters('deniedMcpEndpoints')]", "name": "endpointPatternToolboxes", "where": { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*].serverUrl", "like": "[current('endpointPatternToolboxes')]" } }, "greater": 0 } ] }, { "allOf": [ { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*].type", "equals": "mcp" }, { "field": "Microsoft.CognitiveServices.Data/accounts/projects/toolboxes/tools[*].connectorId", "in": "[parameters('deniedMcpConnectors')]" } ] } ] } }, "greater": 0 } ] } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies Microsoft Foundry agents and toolboxes that use tool types, MCP endpoints or MCP connectors on your denied lists. - **Breaks:** This definition is a preview, and the same definition supports Audit, so a deny only applies when you set the effect to Deny. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DenyAgentAndToolboxTools_Audit.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Prevents Microsoft Foundry agents and toolboxes from using tool types, MCP endpoints, and MCP connectors denied by your organization." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Azure AI Search services should disable public network access (Azure Azure Policy)](/guardrails/azure-policy-search-disable-public-network/) - [Deployment stack deny settings (Azure RBAC deny assignment) (Azure Azure RBAC deny assignments)](/guardrails/azure-deny-assignment-deployment-stack/) - [Resource locks (CanNotDelete and ReadOnly) (Azure Azure resource locks)](/guardrails/azure-lock-cannotdelete-readonly/) ## Sources - Microsoft Foundry agents and toolboxes should not use denied tools (preview): https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Cognitive%20Services/DenyAgentAndToolboxTools_Audit.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-foundry-denied-tools/ --- # Azure AI Search services should disable public network access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure AI Search services should disable public network access is an Azure guardrail written in Azure Policy. Denies Azure AI Search services that leave public network access enabled. What it breaks: existing search services that are public are marked non-compliant but not changed; Microsoft suggests private endpoints to limit exposure. ```text Built-in policy definition: Azure AI Search services should disable public network access Definition ID: ee980b6d-0eca-4501-8d54-f6290fd512c3 Category: Search Supported effects: Audit, Deny, Disabled policyRule (copied from the repository file): { "if": { "allOf": [ { "field": "type", "equals": "Microsoft.Search/searchServices" }, { "field": "Microsoft.Search/searchServices/publicNetworkAccess", "notEquals": "Disabled" } ] }, "then": { "effect": "[parameters('effect')]" } } ``` - **Blocks:** Denies Azure AI Search services that leave public network access enabled. - **Breaks:** Existing search services that are public are marked non-compliant but not changed; Microsoft suggests private endpoints to limit exposure. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - **Test before applying:** As for every Azure Policy guardrail: [how to test before enforcing](/mechanisms/azure-policy/) - **Mechanism:** [Azure Policy (built-in definitions)](/mechanisms/azure-policy/): Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions - **Taken from:** [open source set](https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Search/RequirePublicNetworkAccessDisabled_Deny.json), MIT License, Azure/azure-policy repository, https://raw.githubusercontent.com/Azure/azure-policy/master/LICENSE ## What the source says "Disabling public network access improves security by ensuring that your Azure AI Search service is not exposed on the public internet." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Azure Policy](/mechanisms/azure-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services ## Related pages - [Deployment stack deny settings (Azure RBAC deny assignment) (Azure Azure RBAC deny assignments)](/guardrails/azure-deny-assignment-deployment-stack/) - [Resource locks (CanNotDelete and ReadOnly) (Azure Azure resource locks)](/guardrails/azure-lock-cannotdelete-readonly/) - [Allowed locations assigned at a management group (Azure Management group assignment)](/guardrails/azure-mg-assignment-allowed-locations/) ## Sources - Azure AI Search services should disable public network access: https://raw.githubusercontent.com/Azure/azure-policy/master/built-in-policies/policyDefinitions/Search/RequirePublicNetworkAccessDisabled_Deny.json (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-policy-search-disable-public-network/ --- # Deployment stack deny settings (Azure RBAC deny assignment) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deployment stack deny settings (Azure RBAC deny assignment) is an Azure guardrail written in Azure RBAC deny assignments. A deployment stack created with deny settings makes Azure create a deny assignment that blocks delete (denyDelete) or write and delete (denyWriteAndDelete) on the resources the stack manages, even for users whose role assignments would allow it. ```text Documented deny settings (deployment stacks): DenySettingsMode : Defines the operations that are prohibited on the managed resources to safeguard against unauthorized security principals attempting to delete or update them. This restriction applies to everyone unless you explicitly grant access. The values include: None , DenyDelete , and DenyWriteAndDelete . Azure CLI example from the documentation (resource group scope): az stack group create \ --name '' \ --resource-group '' \ --template-file '' \ --action-on-unmanage 'detachAll' \ --deny-settings-mode 'denyDelete' \ --deny-settings-excluded-actions 'Microsoft.Compute/virtualMachines/write Microsoft.StorageAccounts/delete' \ --deny-settings-excluded-principals ' ' ``` - **Blocks:** A deployment stack created with deny settings makes Azure create a deny assignment that blocks delete (denyDelete) or write and delete (denyWriteAndDelete) on the resources the stack manages, even for users whose role assignments would allow it. - **Breaks:** The deny setting applies only to resources the stack defines explicitly, not to resources created implicitly, and you can exclude at most five principals. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks)) - **Test before applying:** As for every Azure RBAC deny assignments guardrail: [how to test before enforcing](/mechanisms/rbac-deny-assignment/) - **Mechanism:** [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/): Principals named in the deny assignment (all principals when the zero GUID is used), on the managed resources and optionally their child scopes - **Taken from:** [provider documentation](https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments), provider documentation ## What the source says "You can specify deny settings when creating a deployment stack, which creates a deny assignment that the deployment stack resources own." (read 10 Nov 2026). On what it breaks: "The deny setting only applies to explicitly created resources, not implicitly created ones." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity ## Related pages - [Resource locks (CanNotDelete and ReadOnly) (Azure Azure resource locks)](/guardrails/azure-lock-cannotdelete-readonly/) - [Allowed locations assigned at a management group (Azure Management group assignment)](/guardrails/azure-mg-assignment-allowed-locations/) - [Disable service account key creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) ## Sources - Deployment stack deny settings (Azure RBAC deny assignment): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-deny-assignment-deployment-stack/ --- # Resource locks (CanNotDelete and ReadOnly) Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Resource locks (CanNotDelete and ReadOnly) is an Azure guardrail written in Azure resource locks. A CanNotDelete lock stops deletion and a ReadOnly lock stops updates and deletion of a subscription, resource group or resource, overriding user permissions. No documented way to test it before enforcement was found on 10 Nov 2026. ```text CanNotDelete means authorized users can read and modify a resource, but they can't delete it. ReadOnly means authorized users can read a resource, but they can't delete or update it. Applying this lock is similar to restricting all authorized users to the permissions that the Reader role provides. Bicep example from the documentation: resource createRgLock 'Microsoft.Authorization/locks@2016-09-01' = { name: 'rgLock' properties: { level: 'CanNotDelete' notes: 'Resource group should not be deleted.' } } ``` - **Blocks:** A CanNotDelete lock stops deletion and a ReadOnly lock stops updates and deletion of a subscription, resource group or resource, overriding user permissions. - **Breaks:** A read-only lock on a resource group that contains a virtual machine prevents all users from starting or restarting it, because those operations use POST requests. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources)) - **Test before applying:** As for every Azure resource locks guardrail: [how to test before enforcing](/mechanisms/resource-lock/) - **Mechanism:** [Azure resource locks](/mechanisms/resource-lock/): Control plane operations on the locked scope and the resources under it - **Taken from:** [provider documentation](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources), provider documentation ## What the source says "ReadOnly means authorized users can read a resource, but they can't delete or update it." (read 10 Nov 2026). On what it breaks: "A read-only lock on a resource group that contains a virtual machine prevents all users from starting or restarting a virtual machine. These operations require a POST method request." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity ## Related pages - [Allowed locations assigned at a management group (Azure Management group assignment)](/guardrails/azure-mg-assignment-allowed-locations/) - [Disable service account key creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) - [Disable automatic IAM grants for default service accounts (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) ## Sources - Resource locks (CanNotDelete and ReadOnly): https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-lock-cannotdelete-readonly/ --- # Allowed locations assigned at a management group Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allowed locations assigned at a management group is an Azure guardrail written in management group assignment. Assigning the Allowed locations policy at a management group applies it to every subscription and resource under that group through inheritance, so one assignment covers many subscriptions. ```text Management group scope for Azure Policy: assign the built-in definition "Allowed locations" (ID e56962a6-4747-49cd-b67b-bf8b01975c4c, see azure-policy-allowed-locations) at the management group. Microsoft's documented example: "You can create a hierarchy that applies a policy, for example, that limits VM locations to the West US region in the management group called Corp". ``` - **Blocks:** Assigning the Allowed locations policy at a management group applies it to every subscription and resource under that group through inheritance, so one assignment covers many subscriptions. - **Breaks:** Policy and access assigned at a management group cascade to all child subscriptions, and scopes listed in notScopes on the assignment are skipped entirely: they are not evaluated or included in the compliance count. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/scope)) - **Test before applying:** Set enforcementMode to DoNotEnforce on the policy assignment made at the management group, which Microsoft documents as the way to evaluate an assignment before enforcement. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure)) - **Mechanism:** [Management group assignment](/mechanisms/management-groups/): All subscriptions and resources under the management group in the hierarchy - **Taken from:** [provider documentation](https://learn.microsoft.com/en-us/azure/governance/management-groups/overview), provider documentation ## What the source says "You can create a hierarchy that applies a policy, for example, that limits VM locations to the West US region in the management group called Corp" (read 10 Nov 2026). On what it breaks: "Resources within these scopes aren't evaluated or included in the compliance count." ## Related guardrails - [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/): AWS, Service control policies (SCPs), region and residency - [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/): AWS, Service control policies (SCPs), region and residency - [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/): Google Cloud, Organization Policy Service (managed and legacy constraints), region and residency - [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/): Azure, Azure Policy (built-in definitions), region and residency ## Related pages - [Disable service account key creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-sa-key-creation/) - [Disable automatic IAM grants for default service accounts (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) - [Require OS Login (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-os-login/) ## Sources - Allowed locations assigned at a management group: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) - What it breaks: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/scope (read 10 Nov 2026) - How to test it first: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/azure-mg-assignment-allowed-locations/ --- # Disable service account key creation Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Disable service account key creation is a Google Cloud guardrail written in organization Policy Service. Stops users from creating new external service account keys and Cloud Storage HMAC keys in the scope where it is enforced. No documented way to test it before enforcement was found on 10 Nov 2026. ```text constraints/iam.disableServiceAccountKeyCreation This boolean constraint, when enforced, disables the creation of service account external keys and Cloud Storage HMAC keys. By default, service account external keys can be created by users based on their Cloud IAM roles and permissions. This constraint is automatically provisioned as part of Google Cloud security baseline . For more information, see Disable service account key creation . ``` - **Blocks:** Stops users from creating new external service account keys and Cloud Storage HMAC keys in the scope where it is enforced. (legacy managed constraint, boolean.) - **Breaks:** Google documents that user-managed credentials cannot be created for service accounts in affected projects, so tools or workloads that depend on newly created keys cannot get them. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-service-accounts)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, disables the creation of service account external keys and Cloud Storage HMAC keys." (read 10 Nov 2026). On what it breaks: "When this constraint is set, user-managed credentials cannot be created for service accounts in projects affected by the constraint." ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/): Google Cloud, IAM deny policies, identity and privilege ## Related pages - [Disable automatic IAM grants for default service accounts (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/) - [Require OS Login (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-os-login/) - [Deny service account key create and delete to a group on one project (Google Cloud IAM deny policies)](/guardrails/gcp-deny-service-account-key-admin/) ## Sources - Disable service account key creation: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-service-accounts (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-disable-sa-key-creation/ --- # Disable automatic IAM grants for default service accounts Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Disable automatic IAM grants for default service accounts is a Google Cloud guardrail written in organization Policy Service. Stops the default App Engine and Compute Engine service accounts from being automatically granted the Editor role when they are created. No documented way to test it before enforcement was found on 10 Nov 2026. ```text constraints/iam.automaticIamGrantsForDefaultServiceAccounts This boolean constraint, when enforced, prevents the default App Engine and Compute Engine service accounts that are created in your projects from being automatically granted any IAM role on the project when the accounts are created. By default, these service accounts automatically receive the Editor role when they are created. To learn about default service accounts, see https://docs.cloud.google.com/iam/help/service-accounts/default. To learn which roles to grant instead of the Editor role, see https://docs.cloud.google.com/iam/help/service-accounts/troubleshoot-roles-default. This constraint is automatically provisioned as part of Google Cloud security baseline . ``` - **Blocks:** Stops the default App Engine and Compute Engine service accounts from being automatically granted the Editor role when they are created. (legacy managed constraint, boolean.) - **Breaks:** Google documents that it does not stop the Editor or Owner role being granted to default service accounts later, and that other roles must be granted instead of Editor. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-service-accounts)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, prevents the default App Engine and Compute Engine service accounts that are created in your projects from being automatically granted any IAM role on the project when the accounts are created." (read 10 Nov 2026). On what it breaks: "However, it doesn't prevent default service accounts from being granted the Editor or Owner roles later." ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/): Google Cloud, IAM deny policies, identity and privilege ## Related pages - [Require OS Login (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-os-login/) - [Deny service account key create and delete to a group on one project (Google Cloud IAM deny policies)](/guardrails/gcp-deny-service-account-key-admin/) - [Domain restricted sharing (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) ## Sources - Disable automatic IAM grants for default service accounts: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-service-accounts (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/ --- # Require OS Login Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require OS Login is a Google Cloud guardrail written in organization Policy Service. Enables OS Login on all newly created projects and blocks metadata updates that disable OS Login at the project or instance level. No documented way to test it before enforcement was found on 10 Nov 2026. ```text constraints/compute.requireOsLogin This boolean constraint, when enforced, enables OS Login on all newly created Projects. All VM instances created in new projects will have OS Login enabled. On new and existing projects, this constraint prevents metadata updates that disable OS Login at the project or instance level. By default, the OS Login feature is disabled on Compute Engine projects. ``` - **Blocks:** Enables OS Login on all newly created projects and blocks metadata updates that disable OS Login at the project or instance level. (legacy managed constraint, boolean.) - **Breaks:** Google documents that setting OS Login metadata deletes the VM authorized_keys files and VMs no longer accept SSH keys stored in project or instance metadata. ([source](https://cloud.google.com/compute/docs/oslogin/set-up-oslogin)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, enables OS Login on all newly created Projects." (read 10 Nov 2026). On what it breaks: "When you set OS Login metadata, Compute Engine deletes the VM's authorized_keys files and no longer accepts connections from SSH keys that are stored in project or instance metadata." ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/): Google Cloud, IAM deny policies, identity and privilege ## Related pages - [Deny service account key create and delete to a group on one project (Google Cloud IAM deny policies)](/guardrails/gcp-deny-service-account-key-admin/) - [Domain restricted sharing (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) - [Service perimeter around managed services (Google Cloud VPC Service Controls)](/guardrails/gcp-vpcsc-service-perimeter/) ## Sources - Require OS Login: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/compute/docs/oslogin/set-up-oslogin (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-require-os-login/ --- # Deny service account key create and delete to a group on one project Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny service account key create and delete to a group on one project is a Google Cloud guardrail written in iAM deny policies. Denies the serviceAccountKeys.create and serviceAccountKeys.delete permissions to a group on the project or folder where the deny policy is attached, even when the group holds the Service Account Key Admin role. ```json { "deniedPrincipals": [ "principalSet://goog/group/eng@example.com" ], "deniedPermissions": [ "iam.googleapis.com/serviceAccountKeys.create", "iam.googleapis.com/serviceAccountKeys.delete" ] } ``` - **Blocks:** Denies the serviceAccountKeys.create and serviceAccountKeys.delete permissions to a group on the project or folder where the deny policy is attached, even when the group holds the Service Account Key Admin role. - **Breaks:** Google documents that a denied principal cannot use the permission whatever roles they hold, so an exception group (exceptionPrincipals) is needed for anyone who must keep the access. ([source](https://cloud.google.com/iam/docs/deny-overview)) - **Test before applying:** Google documents troubleshooting deny policies with Policy Troubleshooter, which examines the allow, deny and Principal Access Boundary policies that affect a principal; it does not enforce in a preview mode. ([source](https://cloud.google.com/iam/docs/troubleshoot-policies)) - **Mechanism:** [IAM deny policies](/mechanisms/iam-deny/): Principals named in the deny rules, on the resource the policy is attached to and its descendants. - **Taken from:** [provider documentation](https://cloud.google.com/iam/docs/deny-overview), provider documentation ## What the source says "Instead of granting the Service Account Key Admin role on each individual project, you create the following deny rule, which denies create and delete permissions in the Service Account Key Admin role to the principals in the eng group:" (read 10 Nov 2026). On what it breaks: "When a principal is denied a permission, they can't do anything that requires that permission, regardless of the IAM roles they've been granted." ## Related guardrails - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege ## Related pages - [Domain restricted sharing (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-domain-restricted-sharing/) - [Service perimeter around managed services (Google Cloud VPC Service Controls)](/guardrails/gcp-vpcsc-service-perimeter/) - [Allowed destinations for exporting resources (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-allowed-export-destinations/) ## Sources - Deny service account key create and delete to a group on one project: https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) - How to test it first: https://cloud.google.com/iam/docs/troubleshoot-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-deny-service-account-key-admin/ --- # Domain restricted sharing Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Domain restricted sharing is a Google Cloud guardrail written in organization Policy Service. Limits which organization principal sets and Google Workspace customer IDs can be added to IAM policies, so roles cannot be granted to outside identities. No documented way to test it before enforcement was found on 10 Nov 2026. ```yaml name: organizations/ORGANIZATION_ID/policies/iam.allowedPolicyMemberDomains spec: rules: - condition: # This condition applies to the values block. expression: "resource.matchTag('ORGANIZATION_ID/environment', 'dev')" values: allowedValues: - PRINCIPAL_SET - values: allowedValues: - PRINCIPAL_SET ``` - **Blocks:** Limits which organization principal sets and Google Workspace customer IDs can be added to IAM policies, so roles cannot be granted to outside identities. (legacy managed constraint, list.) - **Breaks:** Google documents that the organization principal set and Google Workspace ID are not automatically allowed, so they must be listed to let your own principals be granted roles. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains), provider documentation ## What the source says "Domain-restricted sharing lets you limit resource sharing based on a domain or organization resource." (read 10 Nov 2026). On what it breaks: "Your organization principal set and Google Workspace ID are not automatically allowed." ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/): Google Cloud, Custom organization policy constraints, data exfiltration ## Related pages - [Service perimeter around managed services (Google Cloud VPC Service Controls)](/guardrails/gcp-vpcsc-service-perimeter/) - [Allowed destinations for exporting resources (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-allowed-export-destinations/) - [Custom constraint limiting IAM role grants to organization members (Google Cloud Custom organization policy constraints)](/guardrails/gcp-custom-allow-internal-and-special-identities/) ## Sources - Domain restricted sharing: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-domain-restricted-sharing/ --- # Service perimeter around managed services Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Service perimeter around managed services is a Google Cloud guardrail written in vPC Service Controls. A service perimeter creates a security boundary that blocks, by default, communication from inside the perimeter to Google Cloud services across it, such as copying data to a Cloud Storage bucket outside the perimeter. ```text gcloud access-context-manager perimeters create NAME \ --title=TITLE \ --resources=RESOURCES \ --restricted-services=RESTRICTED-SERVICES \ --ingress-policies=INGRESS-FILENAME.yaml \ --egress-policies=EGRESS-FILENAME.yaml \ [--access-levels=LEVELS] \ [--enable-vpc-accessible-services] \ [--vpc-allowed-services=ACCESSIBLE-SERVICES] \ --policy=POLICY_NAME gcloud access-context-manager perimeters \ dry-run create ProdPerimeter --perimeter-title="Production Perimeter" \ --perimeter-type="regular" \ --perimeter-resources=projects/12345,projects/67890 \ --perimeter-restricted-services=storage.googleapis.com,bigquery.googleapis.com \ --perimeter-ingress-policies=ingress.yaml \ --perimeter-egress-policies=egress.yaml \ --policy=330193482019 ``` - **Blocks:** A service perimeter creates a security boundary that blocks, by default, communication from inside the perimeter to Google Cloud services across it, such as copying data to a Cloud Storage bucket outside the perimeter. - **Breaks:** Google documents that a VM in a VPC network inside the perimeter cannot reach Cloud Storage buckets outside it, and that ingress or egress rules are needed for any cross-perimeter access. ([source](https://cloud.google.com/vpc-service-controls/docs/overview)) - **Test before applying:** Google documents dry run mode for perimeters: requests that violate the perimeter policy are logged and not denied. ([source](https://cloud.google.com/vpc-service-controls/docs/dry-run-mode)) - **Mechanism:** [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/): Google Cloud managed services and the projects in the perimeter; Google says it does not block access to third-party APIs or services on the internet. - **Taken from:** [provider documentation](https://cloud.google.com/vpc-service-controls/docs/create-service-perimeters), provider documentation ## What the source says "Enforced perimeters actively prevent access to protected services." (read 10 Nov 2026). On what it breaks: "VPC Service Controls doesn't allow a VM within a VPC network that is inside a service perimeter to access Cloud Storage buckets that are outside the perimeter." ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/): Google Cloud, Custom organization policy constraints, data exfiltration ## Related pages - [Allowed destinations for exporting resources (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-allowed-export-destinations/) - [Custom constraint limiting IAM role grants to organization members (Google Cloud Custom organization policy constraints)](/guardrails/gcp-custom-allow-internal-and-special-identities/) - [Principal access boundary limiting the resources principals can access (Google Cloud Principal access boundary (PAB) policies)](/guardrails/gcp-pab-eligible-resources/) ## Sources - Service perimeter around managed services: https://cloud.google.com/vpc-service-controls/docs/create-service-perimeters (read 10 Nov 2026) - What it breaks: https://cloud.google.com/vpc-service-controls/docs/overview (read 10 Nov 2026) - How to test it first: https://cloud.google.com/vpc-service-controls/docs/dry-run-mode (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-vpcsc-service-perimeter/ --- # Allowed destinations for exporting resources Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Allowed destinations for exporting resources is a Google Cloud guardrail written in organization Policy Service. Limits which other organizations a resource can be moved to, and denies moves to every other organization. What it breaks: google documents that moves within an organization are not governed by this constraint, and that the move still needs the usual IAM permissions. ```text constraints/resourcemanager.allowedExportDestinations ``` - **Blocks:** Limits which other organizations a resource can be moved to, and denies moves to every other organization. (legacy managed constraint, list.) - **Breaks:** Google documents that moves within an organization are not governed by this constraint, and that the move still needs the usual IAM permissions. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of external Organizations to which resources can be moved, and denies all moves to all other Organizations." (read 10 Nov 2026). On what it breaks: "Moves within an Organization are not governed by this constraint." ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/): Google Cloud, Custom organization policy constraints, data exfiltration ## Related pages - [Custom constraint limiting IAM role grants to organization members (Google Cloud Custom organization policy constraints)](/guardrails/gcp-custom-allow-internal-and-special-identities/) - [Principal access boundary limiting the resources principals can access (Google Cloud Principal access boundary (PAB) policies)](/guardrails/gcp-pab-eligible-resources/) - [Enforce public access prevention on Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) ## Sources - Allowed destinations for exporting resources: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-allowed-export-destinations/ --- # Custom constraint limiting IAM role grants to organization members Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Custom constraint limiting IAM role grants to organization members is a Google Cloud guardrail written in custom organization policy constraints. A custom constraint on iam.googleapis.com/AllowPolicy that permits an IAM allow policy only if every member is in the organization principal set, or is allUsers or allAuthenticatedUsers. ```yaml name: organizations/ORGANIZATION_ID/customConstraints/custom.allowInternalAndSpecialIdentitiesOnly resourceTypes: iam.googleapis.com/AllowPolicy methodTypes: - CREATE - UPDATE condition: "resource.bindings.all( binding, binding.members.all(member, ( MemberInPrincipalSet(member, ['//cloudresourcemanager.googleapis.com/organizations/ORG_ID']) || MemberSubjectMatches(member, ['allUsers', 'allAuthenticatedUsers']) ) ) )" actionType: ALLOW displayName: Only allow organization members, allusers, and allAuthenticatedUsers to be granted roles ``` - **Blocks:** A custom constraint on iam.googleapis.com/AllowPolicy that permits an IAM allow policy only if every member is in the organization principal set, or is allUsers or allAuthenticatedUsers. - **Breaks:** Google documents that with the ALLOW action every case not explicitly listed in the condition is blocked, so any other member type in an allow policy is refused. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains)) - **Test before applying:** Google documents Policy Simulator for Organization Policy, which previews the impact of a new custom constraint before it is enforced, and org policies in dry-run mode. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) - **Mechanism:** [Custom organization policy constraints](/mechanisms/custom-org-policy/): Fields of resource types in services that support custom constraints, in the organization where the constraint is created. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains), provider documentation ## What the source says "For example, the following condition expression restricts role grants to members in your organization, allUsers , and allAuthenticatedUsers :" (read 10 Nov 2026). On what it breaks: "The allow action means that if the condition evaluates to true, the operation to create or update the resource is permitted. This also means that every other case except the one explicitly listed in the condition is blocked." ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration ## Related pages - [Principal access boundary limiting the resources principals can access (Google Cloud Principal access boundary (PAB) policies)](/guardrails/gcp-pab-eligible-resources/) - [Enforce public access prevention on Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) - [Restrict public IP access on Cloud SQL instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) ## Sources - Custom constraint limiting IAM role grants to organization members: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-domains (read 10 Nov 2026) - How to test it first: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-custom-allow-internal-and-special-identities/ --- # Principal access boundary limiting the resources principals can access Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Principal access boundary limiting the resources principals can access is a Google Cloud guardrail written in principal access boundary (PAB) policies. A principal access boundary policy lists the projects, folders and organizations principals are eligible to access, so a principal bound to it is blocked from other resources whatever roles it holds. ```text gcloud iam principal-access-boundary-policies create PAB_POLICY_ID \ --organization=ORG_ID --location=global \ --display-name=DISPLAY_NAME --details-rules=FILE_PATH.json \ --details-enforcement-version=ENFORCEMENT_VERSION { "description": "DESCRIPTION", "resources": [ RESOURCES ], "effect": "ALLOW" } ``` - **Blocks:** A principal access boundary policy lists the projects, folders and organizations principals are eligible to access, so a principal bound to it is blocked from other resources whatever roles it holds. - **Breaks:** Google documents that a policy only blocks permissions in its enforcement version, so an access attempt using a permission it cannot block is not stopped. ([source](https://cloud.google.com/iam/docs/principal-access-boundary-policies)) - **Test before applying:** Google lists Policy Simulator for Principal Access Boundary policies in its documentation, and Policy Troubleshooter examines Principal Access Boundary policies for a given principal, resource and permission. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) - **Mechanism:** [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/): Principal sets bound to the policy through a policy binding; blocks only permissions in the policy enforcement version. - **Taken from:** [provider documentation](https://cloud.google.com/iam/docs/principal-access-boundary-policies-create), provider documentation ## What the source says "A list of Resource Manager resources (projects, folders, and organizations) that you want principals to be eligible to access." (read 10 Nov 2026). On what it breaks: "If a Principal Access Boundary policy can't block a permission, then the policy has no effect on whether principals can use the permission." ## Related guardrails - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration - [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration ## Related pages - [Enforce public access prevention on Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-public-access-prevention/) - [Restrict public IP access on Cloud SQL instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) - [Define allowed external IPs for VM instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vm-external-ip-access/) ## Sources - Principal access boundary limiting the resources principals can access: https://cloud.google.com/iam/docs/principal-access-boundary-policies-create (read 10 Nov 2026) - What it breaks: https://cloud.google.com/iam/docs/principal-access-boundary-policies (read 10 Nov 2026) - How to test it first: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-pab-eligible-resources/ --- # Enforce public access prevention on Cloud Storage Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Enforce public access prevention on Cloud Storage is a Google Cloud guardrail written in organization Policy Service. Blocks ACLs and IAM permissions that grant access to allUsers and allAuthenticatedUsers on Cloud Storage buckets and objects, for existing and future resources. ```text constraints/storage.publicAccessPrevention Secure your Cloud Storage data from public exposure by enforcing public access prevention. This governance policy prevents existing and future resources from being accessed via the public internet by disabling and blocking ACLs and IAM permissions that grant access to allUsers and allAuthenticatedUsers . Enforce this policy on the entire organization (recommended), specific projects, or specific folders to ensure no data is publicly exposed. This policy overrides existing public permissions. Public access will be revoked for existing buckets and objects after this policy is enabled. For more details on the effects of changing enforcement of this constraint on resources, please see: https://docs.cloud.google.com/storage/docs/public-access-prevention. For more information, see Organization policy constraints for Cloud Storage . ``` - **Blocks:** Blocks ACLs and IAM permissions that grant access to allUsers and allAuthenticatedUsers on Cloud Storage buckets and objects, for existing and future resources. (legacy managed constraint, boolean.) - **Breaks:** Google documents that it overrides existing public permissions and revokes public access for existing buckets and objects once enabled. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "Secure your Cloud Storage data from public exposure by enforcing public access prevention." (read 10 Nov 2026). On what it breaks: "This policy overrides existing public permissions. Public access will be revoked for existing buckets and objects after this policy is enabled." ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure ## Related pages - [Restrict public IP access on Cloud SQL instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/) - [Define allowed external IPs for VM instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vm-external-ip-access/) - [Skip default network creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-skip-default-network/) ## Sources - Enforce public access prevention on Cloud Storage: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-storage-public-access-prevention/ --- # Restrict public IP access on Cloud SQL instances Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict public IP access on Cloud SQL instances is a Google Cloud guardrail written in organization Policy Service. Stops Public IP from being configured on Cloud SQL instances where the constraint is enforced. What it breaks: google documents that the constraint is not retroactive: instances with existing Public IP access keep working. ```text constraints/sql.restrictPublicIp This boolean constraint restricts configuring Public IP on Cloud SQL instances where this constraint is enforced. This constraint is not retroactive, Cloud SQL instances with existing Public IP access will still work even after this constraint is enforced. By default, Public IP access is allowed to Cloud SQL instances. ``` - **Blocks:** Stops Public IP from being configured on Cloud SQL instances where the constraint is enforced. (legacy managed constraint, boolean.) - **Breaks:** Google documents that the constraint is not retroactive: instances with existing Public IP access keep working. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint restricts configuring Public IP on Cloud SQL instances where this constraint is enforced." (read 10 Nov 2026). On what it breaks: "This constraint is not retroactive, Cloud SQL instances with existing Public IP access will still work even after this constraint is enforced." ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure ## Related pages - [Define allowed external IPs for VM instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vm-external-ip-access/) - [Skip default network creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-skip-default-network/) - [Restrict VPC peering usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) ## Sources - Restrict public IP access on Cloud SQL instances: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-sql-restrict-public-ip/ --- # Define allowed external IPs for VM instances Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Define allowed external IPs for VM instances is a Google Cloud guardrail written in organization Policy Service. Limits which Compute Engine VM instances can use external IPv4 addresses. What it breaks: google documents that it does not restrict the use of IPv6 addresses. ```text constraints/compute.vmExternalIpAccess This list constraint defines the set of Compute Engine VM instances that are allowed to use external IPv4 addresses. This constraint doesn't restrict the use of IPv6 addresses. By default, all VM instances are allowed to use external IPv4 and IPv6 addresses. The allowed/denied list of VM instances must be identified by the VM instance name, in the form: projects/ PROJECT_ID /zones/ ZONE /instances/ INSTANCE Supported prefix: is: For more information, see Restrict external IP access to specific instances . ``` - **Blocks:** Limits which Compute Engine VM instances can use external IPv4 addresses. (legacy managed constraint, list.) - **Breaks:** Google documents that it does not restrict the use of IPv6 addresses. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of Compute Engine VM instances that are allowed to use external IPv4 addresses." (read 10 Nov 2026). On what it breaks: "This constraint doesn't restrict the use of IPv6 addresses." ## Related guardrails - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure ## Related pages - [Skip default network creation (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-skip-default-network/) - [Restrict VPC peering usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) - [Require predefined policies for VPC flow logs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) ## Sources - Define allowed external IPs for VM instances: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-vm-external-ip-access/ --- # Skip default network creation Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Skip default network creation is a Google Cloud guardrail written in organization Policy Service. Skips creation of the default network and related resources when a project is created where the constraint is enforced. No documented way to test it before enforcement was found on 10 Nov 2026. ```text constraints/compute.skipDefaultNetworkCreation This boolean constraint skips the creation of the default network and related resources during Google Cloud Platform Project resource creation where this constraint is enforced. By default, a default network and supporting resources are automatically created when creating a Project resource. ``` - **Blocks:** Skips creation of the default network and related resources when a project is created where the constraint is enforced. (legacy managed constraint, boolean.) - **Breaks:** Google documents that, by default, a default network and supporting resources are created with each project, so projects created under this constraint have none and need a network created for them. Google also notes that new organization policies are usually not retroactive. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint skips the creation of the default network and related resources during Google Cloud Platform Project resource creation where this constraint is enforced." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Organization Policy Service](/mechanisms/org-policy/). ## Related guardrails - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Restrict VPC peering usage](/guardrails/gcp-orgpolicy-restrict-vpc-peering/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter ## Related pages - [Restrict VPC peering usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-vpc-peering/) - [Require predefined policies for VPC flow logs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) - [Disable Audit Logging exemption (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) ## Sources - Skip default network creation: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-skip-default-network/ --- # Restrict VPC peering usage Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict VPC peering usage is a Google Cloud guardrail written in organization Policy Service. Limits which VPC networks can be peered with the networks of a project, folder or organization. What it breaks: google documents that each peering end is required to have peering permission, so both sides must allow the peer. ```text constraints/compute.restrictVpcPeering This list constraint defines the set of VPC networks that are allowed to be peered with the VPC networks belonging to this project, folder, or organization. Each peering end is required to have peering permission. By default, a Network Admin for one network can peer with any other network. The allowed/denied list of networks must be identified in the form: under:organizations/ ORGANIZATION_ID , under:folders/ FOLDER_ID , under:projects/ PROJECT_ID , or projects/ PROJECT_ID /global/networks/ NETWORK_NAME . Supported prefixes: is: , under: ``` - **Blocks:** Limits which VPC networks can be peered with the networks of a project, folder or organization. (legacy managed constraint, list.) - **Breaks:** Google documents that each peering end is required to have peering permission, so both sides must allow the peer. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of VPC networks that are allowed to be peered with the VPC networks belonging to this project, folder, or organization." (read 10 Nov 2026). On what it breaks: "Each peering end is required to have peering permission." ## Related guardrails - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter ## Related pages - [Require predefined policies for VPC flow logs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/) - [Disable Audit Logging exemption (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) - [Restrict which services may create resources without CMEK (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) ## Sources - Restrict VPC peering usage: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-restrict-vpc-peering/ --- # Require predefined policies for VPC flow logs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Require predefined policies for VPC flow logs is a Google Cloud guardrail written in organization Policy Service. Requires flow logs on all subnetworks in scope, at a minimum sampling rate set by the predefined policy. No documented way to test it before enforcement was found on 10 Nov 2026. ```text constraints/compute.requireVpcFlowLogs This list constraint defines the set of predefined policies that can be enforced for VPC Flow logs. By default VPC Flow logs may be configured with any settings in each subnet. This constraint enforces enabling flow logs for all subnetworks in scope with a required minimum sampling rate. Specify one or more of the following valid values: ESSENTIAL (allows values >= 0.1 and LIGHT (allows values >= 0.5 and COMPREHENSIVE (allows values == 1.0) Supported prefix: is: For more information, see Configure organization policy constraints for VPC Flow Logs . ``` - **Blocks:** Requires flow logs on all subnetworks in scope, at a minimum sampling rate set by the predefined policy. (legacy managed constraint, list.) - **Breaks:** Google documents that subnetworks in scope must have flow logs enabled with the required minimum sampling rate, where by default any settings are allowed. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of predefined policies that can be enforced for VPC Flow logs." (read 10 Nov 2026). On what it breaks: "This constraint enforces enabling flow logs for all subnetworks in scope with a required minimum sampling rate." ## Related guardrails - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Disable Audit Logging exemption](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring ## Related pages - [Disable Audit Logging exemption (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/) - [Restrict which services may create resources without CMEK (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) - [Restrict unencrypted HTTP access to Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) ## Sources - Require predefined policies for VPC flow logs: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-require-vpc-flow-logs/ --- # Disable Audit Logging exemption Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Disable Audit Logging exemption is a Google Cloud guardrail written in organization Policy Service. Prevents exempting additional principals from audit logging. What it breaks: google documents that it does not affect audit-logging exemptions that existed before the constraint was enforced. ```text constraints/iam.disableAuditLoggingExemption This boolean constraint, when enforced, prevents you from exempting additional principals from audit logging. This constraint does not affect any audit-logging exemptions that existed before you enforced the constraint. ``` - **Blocks:** Prevents exempting additional principals from audit logging. (legacy managed constraint, boolean.) - **Breaks:** Google documents that it does not affect audit-logging exemptions that existed before the constraint was enforced. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, prevents you from exempting additional principals from audit logging." (read 10 Nov 2026). On what it breaks: "This constraint does not affect any audit-logging exemptions that existed before you enforced the constraint." ## Related guardrails - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring ## Related pages - [Restrict which services may create resources without CMEK (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/) - [Restrict unencrypted HTTP access to Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) - [Restrict key destruction to disabled key versions (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) ## Sources - Disable Audit Logging exemption: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/ --- # Restrict which services may create resources without CMEK Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict which services may create resources without CMEK is a Google Cloud guardrail written in organization Policy Service. Requires that newly created resources in the listed services are protected by a customer-managed encryption key. What it breaks: google documents that enforcement is not retroactive and that existing non-CMEK resources must be reconfigured or recreated manually. ```text constraints/gcp.restrictNonCmekServices This list constraint defines which services require Customer-Managed Encryption Keys (CMEK). Setting this constraint to Deny (i.e. deny resource creation without CMEK) requires that, for the specified services, newly created resources must be protected by a CMEK key. Supported services that can be set in this constraint are: aiplatform.googleapis.com alloydb.googleapis.com apigee.googleapis.com artifactregistry.googleapis.com backupdr.googleapis.com bigquery.googleapis.com bigquerydatatransfer.googleapis.com bigtable.googleapis.com cloudfunctions.googleapis.com cloudscheduler.googleapis.com cloudtasks.googleapis.com composer.googleapis.com compute.googleapis.com contactcenterinsights.googleapis.com container.googleapis.com dataflow.googleapis.com dataform.googleapis.com datafusion.googleapis.com datamigration.googleapis.com dataplex.googleapis.com dataproc.googleapis.com datastream.googleapis.com datastudio.googleapis.com dialogflow.googleapis.com discoveryengine.googleapis.com documentai.googleapis.com eventarc.googleapis.com file.googleapis.com firestore.googleapis.com gkebackup.googleapis.com integrations.googleapis.com logging.googleapis.com looker.googleapis.com lustre.googleapis.com managedkafka.googleapis.com netapp.googleapis.com notebooks.googleapis.com observability.googleapis.com parametermanager.googleapis.com pubsub.googleapis.com redis.googleapis.com run.googleapis.com secretmanager.googleapis.com securesourcemanager.googleapis.com securitycenter.googleapis.com spanner.googleapis.com speech.googleapis.com sqladmin.googleapis.com storage.googleapis.com storagetransfer.googleapis.com tpu.googleapis.com workstations.googleapis.com Setting this constraint to Deny All is not permitted. Setting this constraint to Allow is not permitted. Enforcement of this constraint is not retroactive. Existing non-CMEK Google Cloud resources must be reconfigured or recreated manually to ensure enforcement. Supported prefix: is: For more information, see CMEK organization policies . ``` - **Blocks:** Requires that newly created resources in the listed services are protected by a customer-managed encryption key. (legacy managed constraint, list.) - **Breaks:** Google documents that enforcement is not retroactive and that existing non-CMEK resources must be reconfigured or recreated manually. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines which services require Customer-Managed Encryption Keys (CMEK)." (read 10 Nov 2026). On what it breaks: "Enforcement of this constraint is not retroactive. Existing non-CMEK Google Cloud resources must be reconfigured or recreated manually to ensure enforcement." ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys ## Related pages - [Restrict unencrypted HTTP access to Cloud Storage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-storage-secure-http-transport/) - [Restrict key destruction to disabled key versions (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) - [Resource location restriction (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-resource-locations/) ## Sources - Restrict which services may create resources without CMEK: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-restrict-non-cmek-services/ --- # Restrict unencrypted HTTP access to Cloud Storage Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict unencrypted HTTP access to Cloud Storage is a Google Cloud guardrail written in organization Policy Service. Explicitly denies HTTP (unencrypted) access to all storage resources. What it breaks: google documents that the Cloud Storage XML API allows unencrypted HTTP by default, so clients using it over HTTP are refused once the constraint is enforced. ```text constraints/storage.secureHttpTransport This boolean constraint, when enforced, explicitly denies HTTP (unencrypted) access to all storage resources. By default, the Cloud Storage XML API allows unencrypted HTTP access. Note that the Cloud Storage JSON API, gRPC, and Cloud console only allow encrypted HTTP access to Cloud Storage resources. ``` - **Blocks:** Explicitly denies HTTP (unencrypted) access to all storage resources. (legacy managed constraint, boolean.) - **Breaks:** Google documents that the Cloud Storage XML API allows unencrypted HTTP by default, so clients using it over HTTP are refused once the constraint is enforced. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, explicitly denies HTTP (unencrypted) access to all storage resources." (read 10 Nov 2026). On what it breaks: "By default, the Cloud Storage XML API allows unencrypted HTTP access." ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys ## Related pages - [Restrict key destruction to disabled key versions (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/) - [Resource location restriction (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-resource-locations/) - [Restrict resource service usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-service-usage/) ## Sources - Restrict unencrypted HTTP access to Cloud Storage: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-storage-secure-http-transport/ --- # Restrict key destruction to disabled key versions Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict key destruction to disabled key versions is a Google Cloud guardrail written in organization Policy Service. Allows destruction only of key versions that are already in the disabled state. What it breaks: google documents that the constraint applies to both new and existing key versions. ```text constraints/cloudkms.disableBeforeDestroy This boolean constraint, when enforced, only allows the destruction of key versions that are in the disabled state. By default, key versions that are in the enabled state and key versions that are in the disabled state can be destroyed. When this constraint is enforced, it applies to both new and existing key versions. ``` - **Blocks:** Allows destruction only of key versions that are already in the disabled state. (legacy managed constraint, boolean.) - **Breaks:** Google documents that the constraint applies to both new and existing key versions. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, only allows the destruction of key versions that are in the disabled state." (read 10 Nov 2026). On what it breaks: "When this constraint is enforced, it applies to both new and existing key versions." ## Related guardrails - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys ## Related pages - [Resource location restriction (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-resource-locations/) - [Restrict resource service usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-service-usage/) - [Restrict Shared VPC host project lien removal (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) ## Sources - Restrict key destruction to disabled key versions: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-kms-disable-before-destroy/ --- # Resource location restriction Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Resource location restriction is a Google Cloud guardrail written in organization Policy Service. Limits the locations where location-based Google Cloud resources can be created. What it breaks: google documents that allowing or denying a multi-region does not imply the same for its sub-locations, so a deny of the us multi-region still allows resources in us-east1. ```text constraints/gcp.resourceLocations This list constraint defines the set of locations where location-based Google Cloud resources can be created. Important: The information on this page does not describe Google Cloud Platform's data location commitments for Customer Data (as defined in the agreement under which Google has agreed to provide Google Cloud Platform services and as described in the Google Cloud Platform Services Summary at https://docs.cloud.google.com/terms/services) to its customers. For the list of Google Cloud Platform services for which Customer Data location may be selected by customers, see Google Cloud Platform Services with Data Residency at https://docs.cloud.google.com/terms/data-residency. By default, resources can be created in any location. For a full list of supported services, see https://docs.cloud.google.com/organization-policy/reference/restrict-locations-supported-services. Policies for this constraint can specify multi-regions such as asia and europe , regions such as us-east1 or europe-west1 as allowed or denied locations. Allowing or denying a multi-region does not imply that all included sub-locations should also be allowed or denied. For example, if the policy denies the us multi-region (which refers to multi-region resources, like some storage services), resources can still be created in the regional location us-east1 . On the other hand, the in:us-locations group contains all locations within the us region, and can be used to block every region. We recommend using value groups to define your policy. You can specify value groups , collections of locations that are curated by Google to provide a simple way to define your resource locations. To use value groups in your organization policy, prefix your entries with the string in: , followed by the value group. For example, to create resources that will only be physically located within the US, set in:us-locations in the list of allowed values. If the suggested_value field is used in a location policy, it should be a region. If the value specified is a region, a UI for a zonal resource may pre-populate any zone in that region. Supported prefixes: is: , in: For more information, see Restrict resource locations . ``` - **Blocks:** Limits the locations where location-based Google Cloud resources can be created. (legacy managed constraint, list.) - **Breaks:** Google documents that allowing or denying a multi-region does not imply the same for its sub-locations, so a deny of the us multi-region still allows resources in us-east1. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of locations where location-based Google Cloud resources can be created." (read 10 Nov 2026). On what it breaks: "Allowing or denying a multi-region does not imply that all included sub-locations should also be allowed or denied." ## Related guardrails - [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/): AWS, Service control policies (SCPs), region and residency - [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/): AWS, Service control policies (SCPs), region and residency - [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/): Azure, Azure Policy (built-in definitions), region and residency ## Related pages - [Restrict resource service usage (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-service-usage/) - [Restrict Shared VPC host project lien removal (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) - [Deny deleting projects tagged prod (Google Cloud IAM deny policies)](/guardrails/gcp-deny-project-delete-prod/) ## Sources - Resource location restriction: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-resource-locations/ --- # Restrict resource service usage Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict resource service usage is a Google Cloud guardrail written in organization Policy Service. Limits which Google Cloud resource services can be used in an organization, folder or project, by allowlist or denylist. What it breaks: google documents that services with a direct dependency on a restricted service are restricted too, for example GKE when Compute Engine is restricted. ```yaml name: organizations/ORGANIZATION_ID/policies/gcp.restrictServiceUsage spec: rules: - values: deniedValues: - file.googleapis.com - bigquery.googleapis.com - storage.googleapis.com ``` - **Blocks:** Limits which Google Cloud resource services can be used in an organization, folder or project, by allowlist or denylist. (legacy managed constraint, list.) - **Breaks:** Google documents that services with a direct dependency on a restricted service are restricted too, for example GKE when Compute Engine is restricted. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-resources)) - **Test before applying:** Google documents that this legacy managed constraint can be used in dry-run mode, where violations are audit logged and not denied. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy)) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/restricting-resources), provider documentation ## What the source says "This constraint can only be enforced on services with resources that are the direct descendants of an organization, folder, or project resource." (read 10 Nov 2026). On what it breaks: "When a service is restricted by this policy, some Google Cloud services that have a direct dependency on the restricted service will be restricted as well." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity - [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/): Google Cloud, IAM deny policies, organization integrity ## Related pages - [Restrict Shared VPC host project lien removal (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/) - [Deny deleting projects tagged prod (Google Cloud IAM deny policies)](/guardrails/gcp-deny-project-delete-prod/) - [Restrict public IP access on new Vertex AI Workbench instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) ## Sources - Restrict resource service usage: https://cloud.google.com/resource-manager/docs/organization-policy/restricting-resources (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-restrict-service-usage/ --- # Restrict Shared VPC host project lien removal Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict Shared VPC host project lien removal is a Google Cloud guardrail written in organization Policy Service. Restricts who can remove a Shared VPC host project lien to users with organization-level permission. What it breaks: google documents that the permission to remove the lien must be granted at the organization level once enforced. ```text constraints/compute.restrictXpnProjectLienRemoval This boolean constraint restricts the set of users that can remove a Shared VPC host project lien without organization-level permission where this constraint is enforced. By default, any user with the permission to update liens can remove a Shared VPC host project lien. Enforcing this constraint requires that permission be granted at the organization level. ``` - **Blocks:** Restricts who can remove a Shared VPC host project lien to users with organization-level permission. (legacy managed constraint, boolean.) - **Breaks:** Google documents that the permission to remove the lien must be granted at the organization level once enforced. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint restricts the set of users that can remove a Shared VPC host project lien without organization-level permission where this constraint is enforced." (read 10 Nov 2026). On what it breaks: "Enforcing this constraint requires that permission be granted at the organization level." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity - [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/): Google Cloud, IAM deny policies, organization integrity ## Related pages - [Deny deleting projects tagged prod (Google Cloud IAM deny policies)](/guardrails/gcp-deny-project-delete-prod/) - [Restrict public IP access on new Vertex AI Workbench instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) - [Disable root access on new Vertex AI Workbench notebooks and instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) ## Sources - Restrict Shared VPC host project lien removal: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/ --- # Deny deleting projects tagged prod Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Deny deleting projects tagged prod is a Google Cloud guardrail written in iAM deny policies. Denies the projects.delete permission to everyone except the project-admins group for projects tagged prod. What it breaks: google documents that the denial condition must use the CEL functions for evaluating resource tags, and other functions and operators are not supported. ```json { "displayName": "Only project admins can delete production projects.", "rules": [ { "denyRule": { "deniedPrincipals": [ "principalSet://goog/public:all" ], "exceptionPrincipals": [ "principalSet://goog/group/project-admins@example.com" ], "deniedPermissions": [ "cloudresourcemanager.googleapis.com/projects.delete" ], "denialCondition": { "title": "Only for prod projects", "expression": "resource.matchTag('12345678/env', 'prod')" } } } ] } ``` - **Blocks:** Denies the projects.delete permission to everyone except the project-admins group for projects tagged prod. - **Breaks:** Google documents that the denial condition must use the CEL functions for evaluating resource tags, and other functions and operators are not supported. ([source](https://cloud.google.com/iam/docs/deny-access)) - **Test before applying:** Google documents troubleshooting deny policies with Policy Troubleshooter, which examines the allow, deny and Principal Access Boundary policies that affect a principal; it does not enforce in a preview mode. ([source](https://cloud.google.com/iam/docs/troubleshoot-policies)) - **Mechanism:** [IAM deny policies](/mechanisms/iam-deny/): Principals named in the deny rules, on the resource the policy is attached to and its descendants. - **Taken from:** [provider documentation](https://cloud.google.com/iam/docs/deny-overview), provider documentation ## What the source says "To solve this problem, you create a deny rule that denies the cloudresourcemanager.googleapis.com/projects.delete permission to everyone except the project-admins group for resources that are tagged prod :" (read 10 Nov 2026). On what it breaks: "Other functions and operators are not supported." ## Related guardrails - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity - [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity - [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity ## Related pages - [Restrict public IP access on new Vertex AI Workbench instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/) - [Disable root access on new Vertex AI Workbench notebooks and instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) - [Define access to models on Vertex AI (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) ## Sources - Deny deleting projects tagged prod: https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) - What it breaks: https://cloud.google.com/iam/docs/deny-access (read 10 Nov 2026) - How to test it first: https://cloud.google.com/iam/docs/troubleshoot-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-deny-project-delete-prod/ --- # Restrict public IP access on new Vertex AI Workbench instances Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict public IP access on new Vertex AI Workbench instances is a Google Cloud guardrail written in organization Policy Service. Restricts public IP access to newly created Vertex AI Workbench notebooks and instances. What it breaks: google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. ```text constraints/ainotebooks.restrictPublicIp This boolean constraint, when enforced, restricts public IP access to newly created Vertex AI Workbench notebooks and instances. By default, public IPs can access Vertex AI Workbench notebooks and instances. ``` - **Blocks:** Restricts public IP access to newly created Vertex AI Workbench notebooks and instances. (legacy managed constraint, boolean.) - **Breaks:** Google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, restricts public IP access to newly created Vertex AI Workbench notebooks and instances." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Organization Policy Service](/mechanisms/org-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services ## Related pages - [Disable root access on new Vertex AI Workbench notebooks and instances (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/) - [Define access to models on Vertex AI (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) - [Disable Grounding with Google Search in generative AI APIs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) ## Sources - Restrict public IP access on new Vertex AI Workbench instances: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/ --- # Disable root access on new Vertex AI Workbench notebooks and instances Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Disable root access on new Vertex AI Workbench notebooks and instances is a Google Cloud guardrail written in organization Policy Service. Prevents newly created Vertex AI Workbench user-managed notebooks and instances from enabling root access. What it breaks: google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. ```text constraints/ainotebooks.disableRootAccess This boolean constraint, when enforced, prevents newly created Vertex AI Workbench user-managed notebooks and instances from enabling root access. By default, Vertex AI Workbench user-managed notebooks and instances can have root access enabled. ``` - **Blocks:** Prevents newly created Vertex AI Workbench user-managed notebooks and instances from enabling root access. (legacy managed constraint, boolean.) - **Breaks:** Google scopes the constraint to newly created notebooks and instances, and states that new organization policies are usually not retroactive. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, prevents newly created Vertex AI Workbench user-managed notebooks and instances from enabling root access." (read 10 Nov 2026). What it breaks is documented for every guardrail of this mechanism: see [Organization Policy Service](/mechanisms/org-policy/). ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services ## Related pages - [Define access to models on Vertex AI (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-allowed-models/) - [Disable Grounding with Google Search in generative AI APIs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) - [Restrict egress domains for data connectors (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) ## Sources - Disable root access on new Vertex AI Workbench notebooks and instances: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - What it breaks: https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/ --- # Define access to models on Vertex AI Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Define access to models on Vertex AI is a Google Cloud guardrail written in organization Policy Service. Defines the set of models and features, such as predict, deploy and tune actions, that can be used in Vertex AI APIs. No documented way to test it before enforcement was found on 10 Nov 2026. ```yaml name: organizations/ORGANIZATION_ID/policies/vertexai.allowedModels spec: rules: values: deniedValues: - publishers/meta/models/llama3:deploy - publishers/google/models/gemini-3.5-flash:tune - publishers/hf-google/models/gemma-2b:deploy ``` - **Blocks:** Defines the set of models and features, such as predict, deploy and tune actions, that can be used in Vertex AI APIs. (legacy managed constraint, list.) - **Breaks:** Google documents that the Model Garden policy only applies to models in Model Garden, not models registered in the Model Registry, and that a policy cannot exceed 500 allowed and denied values. ([source](https://cloud.google.com/vertex-ai/generative-ai/docs/control-model-access)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/vertex-ai/generative-ai/docs/control-model-access), provider documentation ## What the source says "Model Garden's organization policy lets you centrally control the models your users can access and the actions they can take." (read 10 Nov 2026). On what it breaks: "The Model Garden organization policy only applies to models in Model Garden." ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services ## Related pages - [Disable Grounding with Google Search in generative AI APIs (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/) - [Restrict egress domains for data connectors (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) - [Shared reservations owner projects (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) ## Sources - Define access to models on Vertex AI: https://cloud.google.com/vertex-ai/generative-ai/docs/control-model-access (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-vertexai-allowed-models/ --- # Disable Grounding with Google Search in generative AI APIs Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Disable Grounding with Google Search in generative AI APIs is a Google Cloud guardrail written in organization Policy Service. Disables the Grounding with Google Search feature in Vertex AI generative AI APIs. What it breaks: google documents that the feature is enabled by default, so enforcing the constraint removes grounding with Google Search from applications that use it. ```text constraints/vertexai.disableGenAIGoogleSearchGrounding This boolean constraint, when enforced, disables Grounding with Google Search feature in generative AI APIs. By default the feature is enabled. ``` - **Blocks:** Disables the Grounding with Google Search feature in Vertex AI generative AI APIs. (legacy managed constraint, boolean.) - **Breaks:** Google documents that the feature is enabled by default, so enforcing the constraint removes grounding with Google Search from applications that use it. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This boolean constraint, when enforced, disables Grounding with Google Search feature in generative AI APIs." (read 10 Nov 2026). On what it breaks: "By default the feature is enabled." ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services ## Related pages - [Restrict egress domains for data connectors (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/) - [Shared reservations owner projects (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) - [Restrict load balancer creation based on load balancer types (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) ## Sources - Disable Grounding with Google Search in generative AI APIs: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/ --- # Restrict egress domains for data connectors Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict egress domains for data connectors is a Google Cloud guardrail written in organization Policy Service. Allows only URLs with an allowed FQDN to be used by Gemini Enterprise data connectors in projects with VPC Service Controls enabled or in the enforcedProjects list. ```text constraints/discoveryengine.managed.allowedEgressFqdns ``` - **Blocks:** Allows only URLs with an allowed FQDN to be used by Gemini Enterprise data connectors in projects with VPC Service Controls enabled or in the enforcedProjects list. (Managed constraint.) - **Breaks:** Google documents that, when enforced, only URLs with an allowed FQDN can be used for data connectors in the covered projects. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** Google documents that managed constraints support organization policies in dry-run mode, where violations are audit logged and the violating actions are not denied, and that managed constraints support Policy Simulator by default. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This constraint defines the allowed egress fully qualified domain names (FQDNs) for data connectors." (read 10 Nov 2026). On what it breaks: "When enforced, only URLs with an allowed FQDN can be used for data connectors in projects with VPC Service Controls (VPC-SC) enabled, or projects in the enforcedProjects list." ## Related guardrails - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services ## Related pages - [Shared reservations owner projects (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/) - [Restrict load balancer creation based on load balancer types (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) - [Deny member accounts leaving the organization (AWS Service control policies)](/guardrails/aws-scp-deny-leave-org/) ## Sources - Restrict egress domains for data connectors: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/ --- # Shared reservations owner projects Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Shared reservations owner projects is a Google Cloud guardrail written in organization Policy Service. Limits which projects can create and own Compute Engine shared reservations in the organization. What it breaks: google documents the required value format: projects/PROJECT_NUMBER or under:projects/PROJECT_NUMBER. ```text constraints/compute.sharedReservationsOwnerProjects ``` - **Blocks:** Limits which projects can create and own Compute Engine shared reservations in the organization. (legacy managed constraint, list.) - **Breaks:** Google documents the required value format: projects/PROJECT_NUMBER or under:projects/PROJECT_NUMBER. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of projects that are allowed to create and own shared reservations in the org." (read 10 Nov 2026). On what it breaks: "The list of projects allowed to access the shared reservation must be of the form: projects/PROJECT_NUMBER or under:projects/PROJECT_NUMBER ." ## Related guardrails - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Restrict load balancer creation based on load balancer types](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost ## Related pages - [Restrict load balancer creation based on load balancer types (Google Cloud Organization Policy Service)](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/) - [Deny member accounts leaving the organization (AWS Service control policies)](/guardrails/aws-scp-deny-leave-org/) - [Prevent root credentials management in member accounts (AWS Service control policies)](/guardrails/aws-scp-deny-member-root-credentials/) ## Sources - Shared reservations owner projects: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/ --- # Restrict load balancer creation based on load balancer types Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Restrict load balancer creation based on load balancer types is a Google Cloud guardrail written in organization Policy Service. Limits which types of load balancer can be created in an organization, folder or project. What it breaks: google documents that every load balancer type to be allowed or denied must be listed explicitly, from a fixed list of values. ```text constraints/compute.restrictLoadBalancerCreationForTypes This list constraint defines the set of load balancer types which can be created for an organization, folder, or project. Every load balancer type to be allowed or denied must be listed explicitly. By default, creation of all types of load balancers is allowed. The list of allowed or denied values must be identified as the string name of a load balancer, and can only include values from the list below: INTERNAL_TCP_UDP INTERNAL_HTTP_HTTPS GLOBAL_INTERNAL_MANAGED_HTTP_HTTPS GLOBAL_INTERNAL_MANAGED_TCP_PROXY REGIONAL_INTERNAL_MANAGED_TCP_PROXY EXTERNAL_NETWORK_TCP_UDP EXTERNAL_TCP_PROXY EXTERNAL_SSL_PROXY EXTERNAL_HTTP_HTTPS EXTERNAL_MANAGED_HTTP_HTTPS GLOBAL_EXTERNAL_MANAGED_HTTP_HTTPS GLOBAL_EXTERNAL_MANAGED_TCP_PROXY GLOBAL_EXTERNAL_MANAGED_SSL_PROXY REGIONAL_EXTERNAL_MANAGED_TCP_PROXY To include all internal or all external load balancer types, use the in: prefix followed by INTERNAL or EXTERNAL. For example, allowing in:INTERNAL will allow all load balancer types from the above list that include INTERNAL. For more information about restricting load balancer types, see https://docs.cloud.google.com/load-balancing/docs/org-policy-constraints. Supported prefixes: is: , in: For more information, see Organization policy constraints for Cloud Load Balancing . ``` - **Blocks:** Limits which types of load balancer can be created in an organization, folder or project. (legacy managed constraint, list.) - **Breaks:** Google documents that every load balancer type to be allowed or denied must be listed explicitly, from a fixed list of values. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints)) - **Test before applying:** As for every Organization Policy Service guardrail: [how to test before enforcing](/mechanisms/org-policy/) - **Mechanism:** [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/): Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. - **Taken from:** [provider documentation](https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints), provider documentation ## What the source says "This list constraint defines the set of load balancer types which can be created for an organization, folder, or project." (read 10 Nov 2026). On what it breaks: "Every load balancer type to be allowed or denied must be listed explicitly." ## Related guardrails - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost ## Related pages - [Deny member accounts leaving the organization (AWS Service control policies)](/guardrails/aws-scp-deny-leave-org/) - [Prevent root credentials management in member accounts (AWS Service control policies)](/guardrails/aws-scp-deny-member-root-credentials/) - [Deny the root user except for S3 bucket policy changes (AWS Service control policies)](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/) ## Sources - Restrict load balancer creation based on load balancer types: https://cloud.google.com/resource-manager/docs/organization-policy/org-policy-constraints (read 10 Nov 2026) - How to test it first: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/guardrails/gcp-orgpolicy-restrict-lb-creation-types/ --- # 13 preventive mechanisms across three clouds, and where each is evaluated Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Thirteen mechanisms enforce the library's guardrails: four in AWS (SCPs, RCPs, permissions boundaries, declarative policies), four in Azure (Azure Policy, deny assignments, locks, management groups) and five in Google Cloud. Each has its own scope, limits and testing method, and each entry in the library names the one it uses. | Mechanism | Cloud | Applies to | Entries | | --- | --- | --- | --- | | [Service control policies (SCPs)](/mechanisms/scp/) | AWS | IAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. | 27 | | [Resource control policies (RCPs)](/mechanisms/rcp/) | AWS | Resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. | 6 | | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | AWS | A single IAM user or role it is attached to; it does not apply to other principals or to resources. | 2 | | [Declarative policies](/mechanisms/declarative-policy/) | AWS | The configuration of a supported AWS service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. | 5 | | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Azure | Resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions | 27 | | [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/) | Azure | Principals named in the deny assignment (all principals when the zero GUID is used), on the managed resources and optionally their child scopes | 1 | | [Azure resource locks](/mechanisms/resource-lock/) | Azure | Control plane operations on the locked scope and the resources under it | 1 | | [Management group assignment](/mechanisms/management-groups/) | Azure | All subscriptions and resources under the management group in the hierarchy | 1 | | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Google Cloud | Resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. | 25 | | [Custom organization policy constraints](/mechanisms/custom-org-policy/) | Google Cloud | Fields of resource types in services that support custom constraints, in the organization where the constraint is created. | 1 | | [IAM deny policies](/mechanisms/iam-deny/) | Google Cloud | Principals named in the deny rules, on the resource the policy is attached to and its descendants. | 2 | | [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/) | Google Cloud | Google Cloud managed services and the projects in the perimeter; Google says it does not block access to third-party APIs or services on the internet. | 1 | | [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/) | Google Cloud | Principal sets bound to the policy through a policy binding; blocks only permissions in the policy enforcement version. | 1 | **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Platforms by how safely they put a guardrail into production](/platforms/) - [Picks by situation](/situations/) - [Guides](/guides/) ## Sources - Service control policies (SCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - Resource control policies (RCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - IAM permissions boundaries: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Declarative policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) - Azure Policy (built-in definitions): https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - Azure RBAC deny assignments (created by deployment stacks): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - Azure resource locks: https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) - Management group assignment: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/ --- # Service control policies (SCPs): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Service control policies (SCPs) is a AWS mechanism. AWS documents SCPs as an AWS Organizations policy type that sets the maximum available permissions for IAM users and roles in member accounts. It applies to iAM users, IAM roles and the root user in member accounts; not the management account and not service-linked roles. AWS documents SCPs as an AWS Organizations policy type that sets the maximum available permissions for IAM users and roles in member accounts. SCPs grant no permissions; they are evaluated together with identity-based and resource-based policies, and the effective permissions are the intersection. AWS documents an SCP as an AWS Organizations policy type that sets the maximum available permissions for IAM users and roles in member accounts. SCPs grant no permissions; the effective permissions are the intersection of the SCP with identity-based and resource-based policies. They apply to users, roles and the root user in member accounts, and do not affect the management account or service-linked roles. Limits: a maximum document size of 10,240 characters, and at most 10 SCPs attached directly to a root, OU or account, with at least one always attached. To test, the IAM policy simulator evaluates SCPs for identity-based policies but does not surface the matched statements, and AWS recommends trying an SCP on a small set of accounts in a separate OU. 27 library entries use SCPs. ## Documented limits - SCPs do not affect users or roles in the management account. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - SCPs do not affect service-linked roles. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html)) - Maximum SCP document size is 10,240 characters. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html)) - Each root, OU or account can have at most 10 directly attached SCPs, and at least one SCP must always stay attached. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html)) ## Testing before enforcement AWS documents that the IAM policy simulator evaluates SCPs (condition keys and resource scoping in deny statements) for identity-based policies, but it does not surface the matched SCP statements and it does not support RCPs. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) ## The 27 library entries written in it - [Deny member accounts leaving the organization](/guardrails/aws-scp-deny-leave-org/): AWS, Service control policies (SCPs), organization integrity - [Prevent root credentials management in member accounts](/guardrails/aws-scp-deny-member-root-credentials/): AWS, Service control policies (SCPs), identity and privilege - [Deny the root user except for S3 bucket policy changes](/guardrails/aws-scp-deny-root-except-s3-bucket-policy/): AWS, Service control policies (SCPs), identity and privilege - [Deny critical IAM user actions](/guardrails/aws-scp-deny-critical-iam-user-actions/): AWS, Service control policies (SCPs), identity and privilege - [Deny creating or changing IAM federation providers](/guardrails/aws-scp-deny-iam-federation-changes/): AWS, Service control policies (SCPs), identity and privilege - [Prevent creating or expanding public connectivity for VPCs](/guardrails/aws-scp-prevent-vpc-public-connectivity/): AWS, Service control policies (SCPs), network perimeter - [Deny billing, tax and account contact changes](/guardrails/aws-scp-deny-billing-and-contact-changes/): AWS, Service control policies (SCPs), organization integrity - [Deny enabling and disabling AWS Config](/guardrails/aws-scp-deny-config-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny modifications to specific CloudTrail trails](/guardrails/aws-scp-deny-cloudtrail-trail-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny disabling or modifying Amazon GuardDuty](/guardrails/aws-scp-deny-guardduty-changes/): AWS, Service control policies (SCPs), logging and monitoring - [Deny deleting VPC flow logs](/guardrails/aws-scp-deny-vpc-flow-log-deletion/): AWS, Service control policies (SCPs), logging and monitoring - [Deny KMS key deletion](/guardrails/aws-scp-deny-kms-key-deletion/): AWS, Service control policies (SCPs), encryption and keys - [Enforce a 30-day KMS key deletion window](/guardrails/aws-scp-kms-30-day-deletion-window/): AWS, Service control policies (SCPs), encryption and keys - [Deny disabling default EBS encryption](/guardrails/aws-scp-deny-disable-ebs-default-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Require server-side encryption on S3 uploads](/guardrails/aws-scp-require-s3-server-side-encryption/): AWS, Service control policies (SCPs), encryption and keys - [Deny access outside approved Regions](/guardrails/aws-scp-deny-unapproved-regions/): AWS, Service control policies (SCPs), region and residency - [Deny enabling or disabling opt-in Regions](/guardrails/aws-scp-deny-region-enable-disable/): AWS, Service control policies (SCPs), region and residency - [Region deny that allows Bedrock cross-Region inference](/guardrails/aws-scp-region-deny-with-bedrock-cris/): AWS, Service control policies (SCPs), ai services - [Allow only approved Amazon Bedrock models](/guardrails/aws-scp-bedrock-approved-models-only/): AWS, Service control policies (SCPs), ai services - [Deny Amazon Bedrock API keys](/guardrails/aws-scp-deny-bedrock-api-keys/): AWS, Service control policies (SCPs), ai services - [Limit Amazon Bedrock long-term API keys to under 30 days](/guardrails/aws-scp-bedrock-api-keys-max-30-days/): AWS, Service control policies (SCPs), ai services - [Deny IAM and account actions invoked from chat applications](/guardrails/aws-scp-deny-chat-app-iam-actions/): AWS, Service control policies (SCPs), ai services - [Restrict EC2 launches to approved instance types](/guardrails/aws-scp-restrict-ec2-instance-types/): AWS, Service control policies (SCPs), resource and cost - [Restrict AWS Marketplace subscription changes](/guardrails/aws-scp-restrict-marketplace-subscriptions/): AWS, Service control policies (SCPs), resource and cost - [Deny disabling block public access for AMIs](/guardrails/aws-scp-deny-disable-ami-block-public-access/): AWS, Service control policies (SCPs), public exposure - [Deny AWS RAM sharing with external accounts](/guardrails/aws-scp-deny-ram-external-sharing/): AWS, Service control policies (SCPs), data exfiltration - [Require IAM authentication on Lambda function URLs](/guardrails/aws-scp-require-lambda-url-iam-auth/): AWS, Service control policies (SCPs), public exposure **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Resource control policies (RCPs): how it works, its limits and how to test it](/mechanisms/rcp/) - [IAM permissions boundaries: how it works, its limits and how to test it](/mechanisms/permissions-boundary/) - [Declarative policies: how it works, its limits and how to test it](/mechanisms/declarative-policy/) ## Sources - Service control policies (SCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 10 Nov 2026) - Limit: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - Testing: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/scp/ --- # Resource control policies (RCPs): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Resource control policies (RCPs) is a AWS mechanism. AWS documents RCPs as an AWS Organizations policy type that sets the maximum available permissions for resources in member accounts. It applies to resources of supported services in member accounts, for any principal calling them; not resources in the management account and not service-linked roles. AWS documents RCPs as an AWS Organizations policy type that sets the maximum available permissions for resources in member accounts. They apply to requests against those resources regardless of whether the caller belongs to the organization, and they grant no permissions. AWS documents an RCP as an AWS Organizations policy type that sets the maximum available permissions for resources in member accounts. RCPs apply to requests against those resources whether or not the caller belongs to the organization, and grant no permissions. They do not affect resources in the management account or restrict service-linked roles. Limits: a maximum document size of 5,120 characters, and at most 5 RCPs per root, OU or account, the default RCPFullAWSAccess policy included. The IAM policy simulator does not support RCPs, and AWS suggests reviewing CloudTrail logs for Access Denied errors before attaching one. 6 library entries use RCPs, covering controls such as requiring HTTPS and TLS 1.2 for resources. ## Documented limits - RCPs do not affect resources in the management account. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html)) - RCPs do not restrict service-linked roles. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html)) - Maximum RCP document size is 5120 characters. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html)) - At most 5 RCPs can be attached to a root, OU or account, and the default RCPFullAWSAccess policy counts toward that quota. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html)) ## Testing before enforcement AWS documents that the IAM policy simulator does not support RCPs; AWS suggests reviewing CloudTrail logs for Access Denied errors before attaching an RCP. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) ## The 6 library entries written in it - [Require HTTPS for access to resources](/guardrails/aws-rcp-require-https/): AWS, Resource control policies (RCPs), encryption and keys - [Require TLS 1.2 or later for S3](/guardrails/aws-rcp-s3-minimum-tls-1-2/): AWS, Resource control policies (RCPs), encryption and keys - [Deny changes to S3 Block Public Access at bucket level](/guardrails/aws-rcp-deny-bucket-public-access-block-changes/): AWS, Resource control policies (RCPs), public exposure - [Deny unauthenticated access to Cognito identity pools](/guardrails/aws-rcp-deny-cognito-unauthenticated-access/): AWS, Resource control policies (RCPs), public exposure - [Deny S3 customer-provided encryption keys (SSE-C)](/guardrails/aws-rcp-deny-s3-sse-c/): AWS, Resource control policies (RCPs), encryption and keys - [Restrict access to resources in a suspended OU](/guardrails/aws-rcp-restrict-suspended-ou-resources/): AWS, Resource control policies (RCPs), organization integrity **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [IAM permissions boundaries: how it works, its limits and how to test it](/mechanisms/permissions-boundary/) - [Declarative policies: how it works, its limits and how to test it](/mechanisms/declarative-policy/) - [Azure Policy (built-in definitions): how it works, its limits and how to test it](/mechanisms/azure-policy/) ## Sources - Resource control policies (RCPs): https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_rcps.html (read 10 Nov 2026) - Limit: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - Testing: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/rcp/ --- # IAM permissions boundaries: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. IAM permissions boundaries is a AWS mechanism. AWS documents a permissions boundary as a managed policy attached to an IAM user or role that sets the maximum permissions that an identity-based policy can grant to that entity. It applies to a single IAM user or role it is attached to; it does not apply to other principals or to resources. AWS documents a permissions boundary as a managed policy attached to an IAM user or role that sets the maximum permissions that an identity-based policy can grant to that entity. It is evaluated within the account with the entity's other policies. AWS documents a permissions boundary as a managed policy attached to an IAM user or role that sets the maximum permissions an identity-based policy can grant that entity. It limits permissions but grants none, applies only to the one user or role it is attached to, and is supported for users and roles only. A NotPrincipal element with a Deny effect in a resource-based policy always denies any principal that has a boundary attached. The IAM policy simulator can test boundaries, one at a time. The library holds 2 permissions boundary entries; one example limits a user to S3, CloudWatch and EC2 and shows an iam:CreateUser policy failing because the boundary does not allow the action. ## Documented limits - A permissions boundary limits permissions but does not grant them on its own. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html)) - Permissions boundaries are supported for IAM users and roles only. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html)) - A NotPrincipal element with a Deny effect in a resource-based policy always denies any principal that has a permissions boundary attached. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html)) ## Testing before enforcement AWS documents that the IAM policy simulator can test IAM permissions boundaries, one boundary at a time. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) ## The 2 library entries written in it - [Permissions boundary limiting a user to S3, CloudWatch and EC2](/guardrails/aws-pb-restrict-user-to-s3-cloudwatch-ec2/): AWS, IAM permissions boundaries, identity and privilege - [Permissions boundary for delegated IAM user creation](/guardrails/aws-pb-delegated-user-creation-boundary/): AWS, IAM permissions boundaries, identity and privilege **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Declarative policies: how it works, its limits and how to test it](/mechanisms/declarative-policy/) - [Azure Policy (built-in definitions): how it works, its limits and how to test it](/mechanisms/azure-policy/) - [Azure RBAC deny assignments (created by deployment stacks): how it works, its limits and how to test it](/mechanisms/rbac-deny-assignment/) ## Sources - IAM permissions boundaries: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_boundaries.html (read 10 Nov 2026) - Testing: https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/permissions-boundary/ --- # Declarative policies: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Declarative policies is a AWS mechanism. AWS documents declarative policies as an AWS Organizations policy type that declares and enforces a baseline configuration for an AWS service in the service's control plane, rather than regulating API access the way SCPs and RCPs do. AWS documents declarative policies as an AWS Organizations policy type that declares and enforces a baseline configuration for an AWS service in the service's control plane, rather than regulating API access the way SCPs and RCPs do. AWS documents declarative policies as an Organizations policy type that declares and enforces a baseline configuration for an AWS service in the service's control plane, rather than regulating API access the way SCPs and RCPs do. They apply to a supported service attribute in the accounts, OUs or organization they are attached to, including actions by service-linked roles. Limits: a maximum document size of 10,000 characters, and an EC2 declarative policy cannot be changed by account administrators, so any noncompliant action fails. AWS says detaching the policy rolls the attribute back to its previous state. The testing path is an account status report showing the current status of supported attributes. 5 library entries use declarative policies. ## Documented limits - Declarative policies are enforced in the service control plane, not as API authorization. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html)) - Maximum declarative policy document size is 10,000 characters. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html)) - Account administrators cannot change an attribute set by an EC2 declarative policy, and any noncompliant action fails. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html)) - AWS documents that detaching a declarative policy rolls the attribute back to its previous state. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html)) ## Testing before enforcement AWS documents an account status report that shows the current status of the attributes supported by declarative policies for the accounts in scope. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html)) ## The 5 library entries written in it - [Block internet gateway ingress for VPCs (declarative policy)](/guardrails/aws-declarative-vpc-block-public-access/): AWS, Declarative policies, network perimeter - [Block public sharing of EBS snapshots (declarative policy)](/guardrails/aws-declarative-snapshot-block-public-access/): AWS, Declarative policies, data exfiltration - [Require IMDSv2 for new EC2 instances (declarative policy)](/guardrails/aws-declarative-require-imdsv2/): AWS, Declarative policies, identity and privilege - [Enforce an Amazon Bedrock guardrail on model inference (declarative policy)](/guardrails/aws-declarative-bedrock-guardrail-enforcement/): AWS, Declarative policies, ai services - [Opt out of AI service content use organization-wide (declarative policy)](/guardrails/aws-declarative-ai-services-opt-out/): AWS, Declarative policies, ai services **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Azure Policy (built-in definitions): how it works, its limits and how to test it](/mechanisms/azure-policy/) - [Azure RBAC deny assignments (created by deployment stacks): how it works, its limits and how to test it](/mechanisms/rbac-deny-assignment/) - [Azure resource locks: how it works, its limits and how to test it](/mechanisms/resource-lock/) ## Sources - Declarative policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html (read 10 Nov 2026) - Limit: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_reference_limits.html (read 10 Nov 2026) - Limit: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_ec2_syntax.html (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/declarative-policy/ --- # Azure Policy (built-in definitions): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure Policy (built-in definitions) is a Azure mechanism. Azure Policy evaluates resource create and update requests against assigned policy definitions. It applies to resources at the scope of the assignment (management group, subscription, resource group) and their children, except notScopes and exemptions. Azure Policy evaluates resource create and update requests against assigned policy definitions. With the deny effect, a matching request is stopped before it reaches the resource provider; other effects such as modify and deployIfNotExists change or deploy settings instead of blocking. Azure Policy evaluates create and update requests against assigned definitions. With the deny effect a matching request is stopped before it reaches the resource provider; modify and deployIfNotExists change or deploy settings instead. It applies at the scope of the assignment (management group, subscription or resource group) and below, except notScopes and exemptions. Limits: deny acts on create and update, so existing matching resources are only marked noncompliant, and denyAction does not protect child and extension resources when a delete targets the parent. To test, set enforcementMode to DoNotEnforce on the assignment, or assign with the audit effect first. 27 library entries use Azure Policy, taken from the Azure/azure-policy repository. ## Documented limits - Deny acts on create and update requests; existing resources that match are only marked non-compliant. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny)) - Scopes excluded with notScopes are not evaluated at all. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/scope)) - Effects are evaluated in order, and an earlier modify or append can change the request so a deny does not trigger. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-basics)) - denyAction does not protect child and extension resources when the delete targets the parent. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny-action)) ## Testing before enforcement Set enforcementMode to DoNotEnforce on the assignment to evaluate it before enforcement; the audit effect separately creates a warning event without stopping the request. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure)) ## The 27 library entries written in it - [Allowed locations for new resources](/guardrails/azure-policy-allowed-locations/): Azure, Azure Policy (built-in definitions), region and residency - [Allowed resource types](/guardrails/azure-policy-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Not allowed resource types](/guardrails/azure-policy-not-allowed-resource-types/): Azure, Azure Policy (built-in definitions), resource and cost - [Do not allow deletion of resource types](/guardrails/azure-policy-deny-action-delete/): Azure, Azure Policy (built-in definitions), organization integrity - [Allowed virtual machine size SKUs](/guardrails/azure-policy-allowed-vm-sizes/): Azure, Azure Policy (built-in definitions), resource and cost - [Network interfaces should not have public IPs](/guardrails/azure-policy-nic-no-public-ip/): Azure, Azure Policy (built-in definitions), public exposure - [Network interfaces should disable IP forwarding](/guardrails/azure-policy-nic-no-ip-forwarding/): Azure, Azure Policy (built-in definitions), network perimeter - [Gateway subnets should not have a network security group](/guardrails/azure-policy-gateway-subnet-no-nsg/): Azure, Azure Policy (built-in definitions), network perimeter - [Configure storage accounts to disallow public blob access](/guardrails/azure-policy-storage-disable-public-blob/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should disable public network access](/guardrails/azure-policy-storage-disable-public-network/): Azure, Azure Policy (built-in definitions), public exposure - [Storage accounts should have the specified minimum TLS version](/guardrails/azure-policy-storage-min-tls/): Azure, Azure Policy (built-in definitions), encryption and keys - [Storage accounts should prevent shared key access](/guardrails/azure-policy-storage-no-shared-key/): Azure, Azure Policy (built-in definitions), identity and privilege - [Storage SAS tokens should adhere to 7 day maximum validity](/guardrails/azure-policy-storage-sas-7-days/): Azure, Azure Policy (built-in definitions), data exfiltration - [Key vaults should have deletion protection enabled](/guardrails/azure-policy-keyvault-purge-protection/): Azure, Azure Policy (built-in definitions), encryption and keys - [Azure Key Vault should use the RBAC permission model](/guardrails/azure-policy-keyvault-use-rbac/): Azure, Azure Policy (built-in definitions), identity and privilege - [Configure key vaults to enable the firewall](/guardrails/azure-policy-keyvault-firewall/): Azure, Azure Policy (built-in definitions), network perimeter - [OS and data disks should be encrypted with a customer-managed key](/guardrails/azure-policy-disk-cmk/): Azure, Azure Policy (built-in definitions), encryption and keys - [Azure SQL Database should have Microsoft Entra-only authentication enabled](/guardrails/azure-policy-sql-entra-only/): Azure, Azure Policy (built-in definitions), identity and privilege - [Deploy diagnostic settings for Key Vault to Event Hub](/guardrails/azure-policy-keyvault-diagnostic-logs/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Application Insights components should block non-Entra ingestion](/guardrails/azure-policy-appinsights-entra-ingestion/): Azure, Azure Policy (built-in definitions), logging and monitoring - [Configure Cognitive Services accounts to disable public network access](/guardrails/azure-policy-ai-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Configure Cognitive Services accounts to disable local authentication](/guardrails/azure-policy-ai-disable-local-auth/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure AI Services resources to disable local key access](/guardrails/azure-policy-ai-disable-local-auth-dine/): Azure, Azure Policy (built-in definitions), ai services - [Configure Azure Machine Learning workspaces to disable public network access](/guardrails/azure-policy-ml-workspace-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services - [Azure Machine Learning registry deployments restricted to the allowed registry (preview)](/guardrails/azure-policy-ml-registry-models/): Azure, Azure Policy (built-in definitions), ai services - [Microsoft Foundry agents and toolboxes should not use denied tools (preview)](/guardrails/azure-policy-foundry-denied-tools/): Azure, Azure Policy (built-in definitions), ai services - [Azure AI Search services should disable public network access](/guardrails/azure-policy-search-disable-public-network/): Azure, Azure Policy (built-in definitions), ai services **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Azure RBAC deny assignments (created by deployment stacks): how it works, its limits and how to test it](/mechanisms/rbac-deny-assignment/) - [Azure resource locks: how it works, its limits and how to test it](/mechanisms/resource-lock/) - [Management group assignment: how it works, its limits and how to test it](/mechanisms/management-groups/) ## Sources - Azure Policy (built-in definitions): https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny (read 10 Nov 2026) - Limit: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/scope (read 10 Nov 2026) - Limit: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-basics (read 10 Nov 2026) - Limit: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/effect-deny-action (read 10 Nov 2026) - Testing: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/azure-policy/ --- # Azure RBAC deny assignments (created by deployment stacks): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure RBAC deny assignments (created by deployment stacks) is a Azure mechanism. A deny assignment attaches a set of deny actions to a user, group or service principal at a scope. It applies to principals named in the deny assignment (all principals when the zero GUID is used), on the managed resources and optionally their child scopes. A deny assignment attaches a set of deny actions to a user, group or service principal at a scope. Microsoft documents that Azure creates and manages them for generally available scenarios, for example when you set deny settings on a deployment stack. A deny assignment attaches a set of deny actions to a user, group or service principal at a scope. Microsoft documents that Azure creates and manages them for generally available scenarios, for example when deny settings are set on a deployment stack. Customer-managed deny assignments are in private preview. Limits: deployment stack deny settings cover only resources the template defines explicitly, a stack can exclude at most five principals, and deny settings can exclude up to 200 role-based management actions. No simulation or dry run was found for deny assignments on 2026-11-10. 1 library entry uses this mechanism. ## Documented limits - Customer-managed deny assignments are in private preview; for generally available scenarios Azure creates and manages them. ([source](https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments)) - Deployment stack deny settings cover only resources the template defines explicitly. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks)) - A deployment stack can exclude at most five principals from its deny settings. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks)) - Deny settings can exclude up to 200 role-based management actions. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks)) ## Testing before enforcement none found on 2026-11-10 ## The 1 library entry written in it - [Deployment stack deny settings (Azure RBAC deny assignment)](/guardrails/azure-deny-assignment-deployment-stack/): Azure, Azure RBAC deny assignments (created by deployment stacks), organization integrity **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Azure resource locks: how it works, its limits and how to test it](/mechanisms/resource-lock/) - [Management group assignment: how it works, its limits and how to test it](/mechanisms/management-groups/) - [Organization Policy Service (managed and legacy constraints): how it works, its limits and how to test it](/mechanisms/org-policy/) ## Sources - Azure RBAC deny assignments (created by deployment stacks): https://learn.microsoft.com/en-us/azure/role-based-access-control/deny-assignments (read 10 Nov 2026) - Limit: https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/deployment-stacks (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/rbac-deny-assignment/ --- # Azure resource locks: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Azure resource locks is a Azure mechanism. A lock on a subscription, resource group or resource blocks deletion (CanNotDelete) or modification and deletion (ReadOnly) of the locked scope. It applies to control plane operations on the locked scope and the resources under it. A lock on a subscription, resource group or resource blocks deletion (CanNotDelete) or modification and deletion (ReadOnly) of the locked scope. Microsoft documents that the lock overrides any user permissions. A lock on a subscription, resource group or resource blocks deletion (CanNotDelete) or modification and deletion (ReadOnly) of the locked scope, and Microsoft documents that the lock overrides any user permissions. It applies to control plane operations on the scope and the resources under it. Limits: locks do not apply to data plane operations; a ReadOnly lock blocks operations that look like reads but use POST, such as starting a virtual machine; locks cannot be added to management groups; and users with Owner or User Access Administrator roles can create and delete locks. No dry run was found on 2026-11-10. 1 library entry uses resource locks. ## Documented limits - Locks apply only to control plane operations, not data plane operations. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources)) - A ReadOnly lock blocks operations that look like reads but use POST, such as starting a virtual machine. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources)) - Locks cannot be added to management groups. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources)) - Users with Owner or User Access Administrator roles can create and delete locks. ([source](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources)) ## Testing before enforcement none found on 2026-11-10 ## The 1 library entry written in it - [Resource locks (CanNotDelete and ReadOnly)](/guardrails/azure-lock-cannotdelete-readonly/): Azure, Azure resource locks, organization integrity **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Management group assignment: how it works, its limits and how to test it](/mechanisms/management-groups/) - [Organization Policy Service (managed and legacy constraints): how it works, its limits and how to test it](/mechanisms/org-policy/) - [Custom organization policy constraints: how it works, its limits and how to test it](/mechanisms/custom-org-policy/) ## Sources - Azure resource locks: https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/resource-lock/ --- # Management group assignment: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Management group assignment is a Azure mechanism. Management groups organize subscriptions into a hierarchy so that policy and access assigned on a management group cascade to all its subscriptions by inheritance. It applies to all subscriptions and resources under the management group in the hierarchy. Management groups organize subscriptions into a hierarchy so that policy and access assigned on a management group cascade to all its subscriptions by inheritance. Management groups organize subscriptions into a hierarchy, so that policy and access assigned on a management group cascade to all its subscriptions by inheritance. A guardrail assigned there reaches every subscription and resource beneath it. Limits: the tree supports up to six levels of depth, a single directory up to 10,000 management groups, and all subscriptions in one management group must trust the same Microsoft Entra tenant. No dry run was found for the management group itself on 2026-11-10, but a policy assigned there can use enforcementMode DoNotEnforce. 1 library entry covers allowed locations assigned at a management group. ## Documented limits - A management group tree supports up to six levels of depth. ([source](https://learn.microsoft.com/en-us/azure/governance/management-groups/overview)) - A single directory supports up to 10,000 management groups. ([source](https://learn.microsoft.com/en-us/azure/governance/management-groups/overview)) - All subscriptions in one management group must trust the same Microsoft Entra tenant. ([source](https://learn.microsoft.com/en-us/azure/governance/management-groups/overview)) ## Testing before enforcement none found on 2026-11-10 for the management group itself; a policy assigned there can use enforcementMode DoNotEnforce. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure)) ## The 1 library entry written in it - [Allowed locations assigned at a management group](/guardrails/azure-mg-assignment-allowed-locations/): Azure, Management group assignment, region and residency **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Organization Policy Service (managed and legacy constraints): how it works, its limits and how to test it](/mechanisms/org-policy/) - [Custom organization policy constraints: how it works, its limits and how to test it](/mechanisms/custom-org-policy/) - [IAM deny policies: how it works, its limits and how to test it](/mechanisms/iam-deny/) ## Sources - Management group assignment: https://learn.microsoft.com/en-us/azure/governance/management-groups/overview (read 10 Nov 2026) - Testing: https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/management-groups/ --- # Organization Policy Service (managed and legacy constraints): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Organization Policy Service (managed and legacy constraints) is a Google Cloud mechanism. Organization policies set restrictions on Google Cloud resources by enforcing a constraint on an organization, folder or project. It applies to resources in the organization, folder or project where the policy is set and its descendants; Google says application of new policies is usually not retroactive. Organization policies set restrictions on Google Cloud resources by enforcing a constraint on an organization, folder or project. Descendants inherit the policy by default and can override it, subject to the constraint type. Organization policies set restrictions on Google Cloud resources by enforcing a constraint on an organization, folder or project. Descendants inherit the policy by default and can override it, subject to the constraint type. The policy applies to resources in the node where it is set and its descendants. Limits: new organization policies are usually not retroactive, so existing violations are not stopped; managed constraints inherited from a parent are not merged; and dry-run mode is limited to custom constraints, managed constraints and a short list of legacy managed constraints. Policy Simulator cannot test changes to legacy managed constraints. 25 library entries use organization policy constraints. ## Documented limits - New organization policies are usually not retroactive: existing violations are not stopped. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - Dry-run mode is limited to custom constraints, managed constraints and a short list of legacy managed constraints. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy)) - Managed constraints inherited from a parent are not merged to evaluate the effective policy. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) - Policy Simulator cannot test changes to legacy managed constraints. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) ## Testing before enforcement Google documents organization policies in dry-run mode (violations are audit logged, not denied) and Policy Simulator for Organization Policy; legacy managed constraints are mostly excluded from both. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) ## The 25 library entries written in it - [Disable service account key creation](/guardrails/gcp-orgpolicy-disable-sa-key-creation/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Disable automatic IAM grants for default service accounts](/guardrails/gcp-orgpolicy-no-auto-iam-grants-default-sa/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Require OS Login](/guardrails/gcp-orgpolicy-require-os-login/): Google Cloud, Organization Policy Service (managed and legacy constraints), identity and privilege - [Domain restricted sharing](/guardrails/gcp-orgpolicy-domain-restricted-sharing/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Allowed destinations for exporting resources](/guardrails/gcp-orgpolicy-allowed-export-destinations/): Google Cloud, Organization Policy Service (managed and legacy constraints), data exfiltration - [Enforce public access prevention on Cloud Storage](/guardrails/gcp-orgpolicy-storage-public-access-prevention/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Restrict public IP access on Cloud SQL instances](/guardrails/gcp-orgpolicy-sql-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Define allowed external IPs for VM instances](/guardrails/gcp-orgpolicy-vm-external-ip-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), public exposure - [Skip default network creation](/guardrails/gcp-orgpolicy-skip-default-network/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter - [Restrict VPC peering usage](/guardrails/gcp-orgpolicy-restrict-vpc-peering/): Google Cloud, Organization Policy Service (managed and legacy constraints), network perimeter - [Require predefined policies for VPC flow logs](/guardrails/gcp-orgpolicy-require-vpc-flow-logs/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Disable Audit Logging exemption](/guardrails/gcp-orgpolicy-disable-audit-logging-exemption/): Google Cloud, Organization Policy Service (managed and legacy constraints), logging and monitoring - [Restrict which services may create resources without CMEK](/guardrails/gcp-orgpolicy-restrict-non-cmek-services/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Restrict unencrypted HTTP access to Cloud Storage](/guardrails/gcp-orgpolicy-storage-secure-http-transport/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Restrict key destruction to disabled key versions](/guardrails/gcp-orgpolicy-kms-disable-before-destroy/): Google Cloud, Organization Policy Service (managed and legacy constraints), encryption and keys - [Resource location restriction](/guardrails/gcp-orgpolicy-resource-locations/): Google Cloud, Organization Policy Service (managed and legacy constraints), region and residency - [Restrict resource service usage](/guardrails/gcp-orgpolicy-restrict-service-usage/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity - [Restrict Shared VPC host project lien removal](/guardrails/gcp-orgpolicy-restrict-xpn-lien-removal/): Google Cloud, Organization Policy Service (managed and legacy constraints), organization integrity - [Restrict public IP access on new Vertex AI Workbench instances](/guardrails/gcp-orgpolicy-ainotebooks-restrict-public-ip/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable root access on new Vertex AI Workbench notebooks and instances](/guardrails/gcp-orgpolicy-ainotebooks-disable-root-access/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Define access to models on Vertex AI](/guardrails/gcp-orgpolicy-vertexai-allowed-models/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Disable Grounding with Google Search in generative AI APIs](/guardrails/gcp-orgpolicy-vertexai-disable-google-search-grounding/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Restrict egress domains for data connectors](/guardrails/gcp-orgpolicy-discoveryengine-allowed-egress-fqdns/): Google Cloud, Organization Policy Service (managed and legacy constraints), ai services - [Shared reservations owner projects](/guardrails/gcp-orgpolicy-shared-reservations-owner-projects/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost - [Restrict load balancer creation based on load balancer types](/guardrails/gcp-orgpolicy-restrict-lb-creation-types/): Google Cloud, Organization Policy Service (managed and legacy constraints), resource and cost **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Custom organization policy constraints: how it works, its limits and how to test it](/mechanisms/custom-org-policy/) - [IAM deny policies: how it works, its limits and how to test it](/mechanisms/iam-deny/) - [VPC Service Controls (service perimeters): how it works, its limits and how to test it](/mechanisms/vpc-service-controls/) ## Sources - Organization Policy Service (managed and legacy constraints): https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 10 Nov 2026) - Limit: https://cloud.google.com/resource-manager/docs/organization-policy/dry-run-policy (read 10 Nov 2026) - Limit: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/org-policy/ --- # Custom organization policy constraints: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Custom organization policy constraints is a Google Cloud mechanism. A custom constraint is defined in a YAML file that names the resources, methods, conditions and actions subject to the constraint. It applies to fields of resource types in services that support custom constraints, in the organization where the constraint is created. A custom constraint is defined in a YAML file that names the resources, methods, conditions and actions subject to the constraint. Conditions are written in Common Expression Language (CEL), and you then enforce the constraint with an organization policy. A custom constraint is defined in a YAML file that names the resources, methods, conditions and actions subject to the constraint. Conditions are written in Common Expression Language (CEL), and the constraint is then enforced with an organization policy. It applies to fields of resource types in services that support custom constraints. Limits: most resource types support up to 20 custom constraints, a condition can contain up to 1,500 characters, behavior is specific to each service, and a policy set with set-policy takes up to 15 minutes to take effect. Policy Simulator previews the impact of a new custom constraint, and the policy can first be set in dry-run mode. 1 library entry uses a custom constraint. ## Documented limits - Most resource types support up to 20 custom constraints. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints)) - A condition can contain up to 1,500 characters. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints)) - Behavior and supported parameters are specific to each service. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints)) - A policy set with set-policy requires up to 15 minutes to take effect. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints)) ## Testing before enforcement Policy Simulator for Organization Policy previews the impact of a new custom constraint, and the policy can first be set in dry-run mode with the dryRunSpec update mask. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) ## The 1 library entry written in it - [Custom constraint limiting IAM role grants to organization members](/guardrails/gcp-custom-allow-internal-and-special-identities/): Google Cloud, Custom organization policy constraints, data exfiltration **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [IAM deny policies: how it works, its limits and how to test it](/mechanisms/iam-deny/) - [VPC Service Controls (service perimeters): how it works, its limits and how to test it](/mechanisms/vpc-service-controls/) - [Principal access boundary (PAB) policies: how it works, its limits and how to test it](/mechanisms/principal-access-boundary/) ## Sources - Custom organization policy constraints: https://cloud.google.com/resource-manager/docs/organization-policy/creating-managing-custom-constraints (read 10 Nov 2026) - Testing: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/custom-org-policy/ --- # IAM deny policies: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. IAM deny policies is a Google Cloud mechanism. Deny policies are made up of deny rules that prevent certain principals from using certain permissions, regardless of the roles they are granted. It applies to principals named in the deny rules, on the resource the policy is attached to and its descendants. Deny policies are made up of deny rules that prevent certain principals from using certain permissions, regardless of the roles they are granted. IAM checks relevant deny policies before allow policies, and a deny policy is attached to a project, folder or organization. Deny policies are made of deny rules that prevent certain principals from using certain permissions, regardless of the roles they hold. IAM checks relevant deny policies before allow policies, and a deny policy is attached to a project, folder or organization. Limits: each resource can have up to 500 deny policies, the policies on a resource can contain 500 deny rules in total, and wildcards are supported only in supported permission groups. Policy Troubleshooter examines the allow, deny and principal access boundary policies that affect a principal and says whether it can use a permission on a resource. 2 library entries use IAM deny policies. ## Documented limits - Each resource can have up to 500 deny policies. ([source](https://cloud.google.com/iam/docs/deny-overview)) - The deny policies on a resource can contain a total of 500 deny rules. ([source](https://cloud.google.com/iam/docs/deny-overview)) - Wildcards are supported only in supported permission groups. ([source](https://cloud.google.com/iam/docs/deny-overview)) ## Testing before enforcement Google documents Policy Troubleshooter, which examines the allow, deny and Principal Access Boundary policies that affect a principal and says whether the principal can use a permission on a resource. ([source](https://cloud.google.com/iam/docs/troubleshoot-policies)) ## The 2 library entries written in it - [Deny service account key create and delete to a group on one project](/guardrails/gcp-deny-service-account-key-admin/): Google Cloud, IAM deny policies, identity and privilege - [Deny deleting projects tagged prod](/guardrails/gcp-deny-project-delete-prod/): Google Cloud, IAM deny policies, organization integrity **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [VPC Service Controls (service perimeters): how it works, its limits and how to test it](/mechanisms/vpc-service-controls/) - [Principal access boundary (PAB) policies: how it works, its limits and how to test it](/mechanisms/principal-access-boundary/) - [Service control policies (SCPs): how it works, its limits and how to test it](/mechanisms/scp/) ## Sources - IAM deny policies: https://cloud.google.com/iam/docs/deny-overview (read 10 Nov 2026) - Testing: https://cloud.google.com/iam/docs/troubleshoot-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/iam-deny/ --- # VPC Service Controls (service perimeters): what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. VPC Service Controls (service perimeters) is a Google Cloud mechanism. A service perimeter creates a security boundary around Google Cloud resources. It applies to google Cloud managed services and the projects in the perimeter; Google says it does not block access to third-party APIs or services on the internet. A service perimeter creates a security boundary around Google Cloud resources. It allows free communication within the perimeter but, by default, blocks communication to Google Cloud services across it. A service perimeter creates a security boundary around Google Cloud resources. It allows free communication within the perimeter but, by default, blocks communication to Google Cloud services across it. It applies to Google Cloud managed services and the projects in the perimeter. Limits: not all Google Cloud services are supported, the perimeter does not block third-party APIs or services on the internet, and changes can take up to 30 minutes to propagate and may briefly block requests. Dry run mode logs requests that violate the perimeter policy without denying them. 1 library entry, a service perimeter around managed services, uses it. ## Documented limits - Not all Google Cloud services are supported. ([source](https://cloud.google.com/vpc-service-controls/docs/overview)) - The perimeter does not block third-party APIs or services on the internet. ([source](https://cloud.google.com/vpc-service-controls/docs/overview)) - Changes can take up to 30 minutes to propagate and may briefly block requests. ([source](https://cloud.google.com/vpc-service-controls/docs/create-service-perimeters)) ## Testing before enforcement Dry run mode logs requests that violate the perimeter policy without denying them. ([source](https://cloud.google.com/vpc-service-controls/docs/dry-run-mode)) ## The 1 library entry written in it - [Service perimeter around managed services](/guardrails/gcp-vpcsc-service-perimeter/): Google Cloud, VPC Service Controls (service perimeters), data exfiltration **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Principal access boundary (PAB) policies: how it works, its limits and how to test it](/mechanisms/principal-access-boundary/) - [Service control policies (SCPs): how it works, its limits and how to test it](/mechanisms/scp/) - [Resource control policies (RCPs): how it works, its limits and how to test it](/mechanisms/rcp/) ## Sources - VPC Service Controls (service perimeters): https://cloud.google.com/vpc-service-controls/docs/overview (read 10 Nov 2026) - Limit: https://cloud.google.com/vpc-service-controls/docs/create-service-perimeters (read 10 Nov 2026) - Testing: https://cloud.google.com/vpc-service-controls/docs/dry-run-mode (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/vpc-service-controls/ --- # Principal access boundary (PAB) policies: what it can block and where it stops Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. Principal access boundary (PAB) policies is a Google Cloud mechanism. A principal access boundary policy defines the resources a set of principals is eligible to access. It applies to principal sets bound to the policy through a policy binding; blocks only permissions in the policy enforcement version. A principal access boundary policy defines the resources a set of principals is eligible to access. If a principal is ineligible to access a resource, access is limited regardless of the roles granted; the policy can only block access, not grant it. A principal access boundary policy defines the resources a set of principals is eligible to access. If a principal is ineligible for a resource, access is limited regardless of the roles granted. The policy can only block access, never grant it, and applies to principal sets bound through a policy binding. Limits: a policy can have up to 500 rules, an organization up to 1,000 policies, and permissions outside the policy enforcement version are not blocked. Cached publicly visible resources can still be viewed. Google lists Policy Simulator for these policies in its documentation, and Policy Troubleshooter examines them. 1 library entry uses a principal access boundary. ## Documented limits - A policy can have up to 500 rules. ([source](https://cloud.google.com/iam/docs/principal-access-boundary-policies)) - An organization can have up to 1000 policies. ([source](https://cloud.google.com/iam/docs/principal-access-boundary-policies)) - Permissions outside the enforcement version are not blocked. ([source](https://cloud.google.com/iam/docs/principal-access-boundary-policies)) - Cached publicly visible resources can still be viewed. ([source](https://cloud.google.com/iam/docs/principal-access-boundary-policies)) ## Testing before enforcement Google lists Policy Simulator for Principal Access Boundary policies in its documentation navigation, and Policy Troubleshooter examines PAB policies. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) ## The 1 library entry written in it - [Principal access boundary limiting the resources principals can access](/guardrails/gcp-pab-eligible-resources/): Google Cloud, Principal access boundary (PAB) policies, data exfiltration **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [Service control policies (SCPs): how it works, its limits and how to test it](/mechanisms/scp/) - [Resource control policies (RCPs): how it works, its limits and how to test it](/mechanisms/rcp/) - [IAM permissions boundaries: how it works, its limits and how to test it](/mechanisms/permissions-boundary/) ## Sources - Principal access boundary (PAB) policies: https://cloud.google.com/iam/docs/principal-access-boundary-policies (read 10 Nov 2026) - Testing: https://cloud.google.com/policy-intelligence/docs/test-organization-policies (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/mechanisms/principal-access-boundary/ --- # The Simulation Coverage Matrix: which platforms show a guardrail's effect before it ships Facts read 10 Nov 2026. Page reviewed 16 Nov 2026. The matrix records four checks for each of sixteen platforms: whether a change is simulated before it applies, whether the simulation runs against the real environment, whether the impact is shown, and whether a native dry run exists. Each cell rests on a fact read on 10 Nov 2026, or says none found. The four columns ask different questions. Simulated before apply asks whether the product tests a guardrail before it takes effect. Against the real environment asks whether that test uses the customer's own configuration or logs, not an example. Impact shown asks whether the output says what would be blocked. Native dry run asks whether the cloud's own mechanism has a mode that logs violations without denying, such as Azure Policy's DoNotEnforce or Google Cloud's organization policy dry run. Read the matrix by cell, not by row. A cell is filled only when a page says so; where a page was read and nothing was found, the cell says "none found on 2026-11-10". A preview of a generated script is not a simulation of impact, and a validate button is not a test against the environment. Vendor claims are the vendor's own: "Act says" is not the same as a result. ## The 16 platforms | Platform | Safe to enforce (0 to 10) | What the evidence says about testing first | Documented limit | | --- | --- | --- | --- | | [Act Security](/platforms/act/) | 7 | "Simulation: Test changes against a digital twin to see their exact impact before enforcing, ensuring nothing breaks." ([read 10 Nov 2026](https://act.security/platform?utm_source=cloudgovernancetools.com&utm_medium=referral&utm_campaign=asset)) | "The main areas to validate are access-model accuracy, simulation coverage for rarely used or undocumented dependencies, provider and service depth" | | [Sonrai Security](/platforms/sonrai/) | 5 | "Policies are built from your usage data. Anything in use is exempted before it’s written. You see the exemption list first." ([read 10 Nov 2026](https://sonraisecurity.com)) | none found | | [AWS Control Tower](/platforms/aws-control-tower/) | 1 | none found on 10 Nov 2026 | "We recommend that you test proactive controls individually before you enable them in your environment." | | [Google Cloud IAM recommender](/platforms/google-iam-recommender/) | 6 | "Policy Simulator for Identity and Access Management allow policies lets you see how a change to an allow policy might affect a principal's access before you commit to making the change." ([read 10 Nov 2026](https://cloud.google.com/policy-intelligence/docs/iam-simulator-overview)) | "Each simulator only evaluates one policy type" | | [Turbot](/platforms/turbot/) | 2 | "have a policy/ collection of policies configured such that Guardrails is only monitoring cloud resources rather than remediating. Controls in alarm should be investigated by administrators, as enforcement requires review." ([read 10 Nov 2026](https://turbot.com/guardrails/docs/concepts/controls)) | "Enforcement actions will not run." | | [Apono](/platforms/apono/) | 3 | "Previews the proposed change. Apono applies only changes that have been previewed. The preview summarizes who receives what access, for how long, and who approves it. It also warns if the change expands access." ([read 10 Nov 2026](https://docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md)) | "On the Quarantine Principal tab, click Remediate to limit access within the dedicated access flow." | | [Britive](/platforms/britive/) | 1 | "Enter the Name and the Policy code in JSON format in the Create Policy page. Click Validate to validate the policy details." ([read 10 Nov 2026](https://docs.britive.com/docs/profilemgmt-managing-profiles.md)) | "This discrepancy is resolved during the next profile checkout." | | [AWS IAM Access Analyzer](/platforms/aws-iam-access-analyzer/) | 4 | "policy affects public and cross-account access to your resource before deploying resource" ([read 10 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html)) | "IAM Access Analyzer does not identify action-level activity for data events, such as Amazon S3 data events, in generated policies." | | [Wiz](/platforms/wiz/) | 4 | "The Green Agent can use Security Graph context and historical activity to evaluate whether a proposed remediation could disrupt an underlying application." ([read 10 Nov 2026](https://softwareanalyst.io/reports/the-ciso-s-guide-to-cloud-security-intent-enforcement)) | "although historical usage cannot establish every infrequent or future dependency." | | [P0 Security](/platforms/p0/) | 0 | none found on 10 Nov 2026 | none found | | [CrowdStrike](/platforms/crowdstrike/) | 4 | "One-click remediation testing: Simulate remediation tactics to understand outcomes and ensure confidence by performing a dry run prior to deployment." ([read 10 Nov 2026](https://www.crowdstrike.com/blog/crowdstrike-extends-ciem-capabilities/)) | none found | | [Tenable Cloud Security](/platforms/tenable/) | 3 | Tenable's blog says the product generates right-sized policies the user can apply and shows the exact difference between the generated policy and the current policy. The page does not describe a simulation against the environment. ([read 10 Nov 2026](https://www.tenable.com/blog/least-privilege-policy-automated-analysis-trumps-native-aws-tools)) | none found | | [Entitle](/platforms/entitle/) | 0 | none found on 10 Nov 2026 | none found | | [Sysdig](/platforms/sysdig/) | 2 | "With Sysdig’s careful recommendations, teams can confidently tighten IAM policies without disrupting legitimate activity." ([read 10 Nov 2026](https://www.sysdig.com/solutions/cloud-infrastructure-entitlement-management-ciem)) | none found | | [Prisma Cloud](/platforms/prisma/) | 0 | none found on 10 Nov 2026 | none found | | [Microsoft](/platforms/microsoft/) | 0 | none found on 10 Nov 2026 | "Select Mode , and then, if you want recommendations to be generated and applied manually, select On-Demand ." | Ranked view order. The score is the safe to enforce criterion of this site's rubric; quotes are each vendor's own material unless marked. ## What each cloud offers natively | Mechanism | Cloud | Test before enforcement | | --- | --- | --- | | [Service control policies (SCPs)](/mechanisms/scp/) | AWS | AWS documents that the IAM policy simulator evaluates SCPs (condition keys and resource scoping in deny statements) for identity-based policies, but it does not surface the matched SCP statements and it does not support RCPs. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) | | [Resource control policies (RCPs)](/mechanisms/rcp/) | AWS | AWS documents that the IAM policy simulator does not support RCPs; AWS suggests reviewing CloudTrail logs for Access Denied errors before attaching an RCP. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) | | [IAM permissions boundaries](/mechanisms/permissions-boundary/) | AWS | AWS documents that the IAM policy simulator can test IAM permissions boundaries, one boundary at a time. ([source](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_testing-policies.html)) | | [Declarative policies](/mechanisms/declarative-policy/) | AWS | AWS documents an account status report that shows the current status of the attributes supported by declarative policies for the accounts in scope. ([source](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_declarative.html)) | | [Azure Policy (built-in definitions)](/mechanisms/azure-policy/) | Azure | Set enforcementMode to DoNotEnforce on the assignment to evaluate it before enforcement; the audit effect separately creates a warning event without stopping the request. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure)) | | [Azure RBAC deny assignments (created by deployment stacks)](/mechanisms/rbac-deny-assignment/) | Azure | none found on 2026-11-10 | | [Azure resource locks](/mechanisms/resource-lock/) | Azure | none found on 2026-11-10 | | [Management group assignment](/mechanisms/management-groups/) | Azure | none found on 2026-11-10 for the management group itself; a policy assigned there can use enforcementMode DoNotEnforce. ([source](https://learn.microsoft.com/en-us/azure/governance/policy/concepts/assignment-structure)) | | [Organization Policy Service (managed and legacy constraints)](/mechanisms/org-policy/) | Google Cloud | Google documents organization policies in dry-run mode (violations are audit logged, not denied) and Policy Simulator for Organization Policy; legacy managed constraints are mostly excluded from both. ([source](https://cloud.google.com/resource-manager/docs/organization-policy/overview)) | | [Custom organization policy constraints](/mechanisms/custom-org-policy/) | Google Cloud | Policy Simulator for Organization Policy previews the impact of a new custom constraint, and the policy can first be set in dry-run mode with the dryRunSpec update mask. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) | | [IAM deny policies](/mechanisms/iam-deny/) | Google Cloud | Google documents Policy Troubleshooter, which examines the allow, deny and Principal Access Boundary policies that affect a principal and says whether the principal can use a permission on a resource. ([source](https://cloud.google.com/iam/docs/troubleshoot-policies)) | | [VPC Service Controls (service perimeters)](/mechanisms/vpc-service-controls/) | Google Cloud | Dry run mode logs requests that violate the perimeter policy without denying them. ([source](https://cloud.google.com/vpc-service-controls/docs/dry-run-mode)) | | [Principal access boundary (PAB) policies](/mechanisms/principal-access-boundary/) | Google Cloud | Google lists Policy Simulator for Principal Access Boundary policies in its documentation navigation, and Policy Troubleshooter examines PAB policies. ([source](https://cloud.google.com/policy-intelligence/docs/test-organization-policies)) | ## Cite this **Cite this.** Cloud Governance Tools, Simulation Coverage Matrix, edition 10 Nov 2026, https://cloudgovernancetools.com/simulation/ **Where Act is behind.** Act writes guardrails in the provider's own mechanisms and hands them to your pipeline to enforce, so it is bounded by what those mechanisms can express in each cloud. It does not apply them itself, the boundary types it names are AWS's and Azure's, and exception and waiver handling after a guardrail ships was described as roadmap. ## Related pages - [AWS guardrails: 40 preventive policies with what they block and break](/library/aws/) - [Azure guardrails: 30 preventive policies with what they block and break](/library/azure/) - [Google Cloud guardrails: 30 preventive policies with what they block and break](/library/gcp/) ## Sources - act.security/platform: https://act.security/platform (read 10 Nov 2026) - sonraisecurity.com: https://sonraisecurity.com (read 10 Nov 2026) - docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html: https://docs.aws.amazon.com/controltower/latest/userguide/proactive-controls.html (read 10 Nov 2026) - cloud.google.com/policy-intelligence/docs/iam-simulator-overview: https://cloud.google.com/policy-intelligence/docs/iam-simulator-overview (read 10 Nov 2026) - turbot.com/guardrails/docs/concepts/controls: https://turbot.com/guardrails/docs/concepts/controls (read 10 Nov 2026) - docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md: https://docs.apono.io/docs/audits-and-reports/apono-admin-mcp-server.md (read 10 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudgovernancetools.com/simulation/