AIF-C01 Practice Question: Security, Compliance, and Governance for AI Solutions

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  S3 bucket policy on the model artifacts

  Why it's wrong here

  S3 bucket policies govern access to objects stored in a bucket, not invocation of a Bedrock base model, whose authorisation is enforced by Bedrock resource policies. They are tempting because model artefacts and custom models often reside in S3, so a bucket policy would be correct if the requirement were protecting those stored artefacts.

• ✓

  Bedrock resource-based policy

  Why this is correct

  Bedrock resource-based policies attach directly to a specific base model, letting you define which principals may invoke it and under what conditions. This satisfies the stem's constraint of restricting IAM roles per model, since identity-based policies alone cannot scope permissions to an individual Bedrock model resource.

• ✗

  AWS KMS key policy for the encryption key

  Why it's wrong here

  A KMS key policy governs who may use a key for cryptographic operations, not who may invoke a Bedrock model; Bedrock authorisation flows through IAM resource policies on the model. It is tempting because KMS policies do control cross-account access to encrypted resources, which would be the right tool if the requirement concerned key usage rather than model invocation.

• ✗

  AWS CloudTrail log delivery policy

  Why it's wrong here

  CloudTrail log delivery policies control which S3 buckets or CloudWatch Logs groups receive trail events; they cannot authorise or deny Bedrock model invocation. They are tempting because CloudTrail records who invoked a model, so it would be the correct choice if the requirement were auditing invocations rather than restricting them.

Go deeper

Related to this question