AI-Assisted Secure Coding (Appendix C) · AI-Generated Infrastructure & Pipeline Artifacts
Policy-check AI infrastructure changes
Verify that AI-generated infrastructure and workflow changes pass policy-as-code enforcement (OPA, Conftest, Checkov, tfsec, KICS, kube-linter) at the same level as, or stricter than, human-authored changes. Policy violations block promotion.
OWASP AISVS v1.01, requirement AC.7.3
In plain English
Run policy-as-code checks on AI-generated infrastructure and workflow changes, at least as strictly as on changes people write, and block promotion when a policy is broken. Automated policy catches the classic mistakes: public storage buckets, wildcard permissions, open security groups.
What good looks like
- Policy checks run on every infrastructure and workflow change.
- The rules for AI-generated changes are the same or stricter.
- A violation blocks promotion.
- The policy set is versioned and reviewed.
How teams usually do it
- Tools such as Open Policy Agent with Conftest, Checkov, Trivy (which has absorbed tfsec), KICS or kube-linter in CI.
- The same policies enforced again at deploy time by an admission controller or cloud policy service.
- A workflow linter such as zizmor or actionlint for CI definitions.
Suggested evidence
- CI configuration.
- A blocked change.
- The policy repository history.
Common mistakes
- Policy checks set to warn only.
- Inline suppressions added by the same agent that wrote the change.
- Checking Terraform and forgetting workflow files and Kubernetes manifests.
Not sure which requirements apply to your system? Take the assessment, or browse every guide in this chapter.
Requirement text from the OWASP AI Security Verification Standard, licensed under CC BY-SA 4.0. This guidance is shared under the same licence.