#ai-governance

AI Safety should have a starter file audit before this source set is treated as finished. The attached AI Safety source review worksheet is the working document: fill it out against AI RMF Playbook, then keep or reject the claim in the ai safety playbooks folio.

The audit fields are reader problem, source checked, claim to verify, decision affected, missing context, and next test. They force the post to name the user problem, the checked source, and the next test before anyone saves the idea as a template, tracker, checklist, or worksheet.

AI RMF Playbook anchors the audit. The practical check is whether it changes a concrete AI safety decision, checklist item, tracker row, worksheet field, or template example.

NIST AI Risk Management Framework is the cross-check. If it changes the conclusion, the worksheet should preserve the disagreement instead of hiding it in a clean paragraph.

Inspect source is the artifact source. It should give the post a document, dataset, repository, guide, API, checklist, or worked-example angle that a reader can reuse.

Video cross-check: Anthropic video archive should be used for workflow texture, not as the only citation. Watch for the exact screen, demo, route, pattern, cut setting, session step, or process moment that would change the audit.

Two sources to open first: airc.nist.gov/AI_RMF_Knowledge_Base/Playbook and youtube.com/@AnthropicAI/videos. File the note under ai-safety, red-teaming, model-evals, ai-governance, then use the attached document to record which claim each source supports, which claim remains opinion, and which detail should be removed if nobody can verify it.

The strongest reply would improve one field in the document. A better source, clearer caveat, stronger example, or missing beginner trap is more valuable than a broad reaction.