Skip to content
Cloud Governance Tools

Enforce an Amazon Bedrock guardrail on model inference (declarative policy)

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.

More ai services guardrails ยท About Declarative policies

{
  "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)
Test before applying
As for every Declarative policies guardrail: how to test before enforcing
Mechanism
Declarative policies: 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, 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."

Sources

  1. Enforce an Amazon Bedrock guardrail on model inference (declarative policy), read 10 Nov 2026

Last reviewed 16 Nov 2026.