policy-monitor

AI policy drift occurs when internal deployments change without updating governance documents. You detect it by diffing saved AI outputs, triage results, and vendor reviews against current policy commitments and the use case registry.

To check if a proposed AI use case is covered, diff the new practice against current policy commitments and the use case registry. This outputs covered, missing, and conflicting areas with draft language and timing guidance.

Run a weekly drift sweep by reading configured local outputs like vendor AI reviews and triage results, then diffing newly saved activity against current policy commitments to identify gaps requiring updates.

Yes, AI policy monitoring requires a use case registry. The Skill reads the local configuration state and authoritative policy document, then diffs activity against the registry to produce REQUIRED or ADVISABLE gap findings.

REQUIRED findings identify policy contradictions or misrepresentations, while ADVISABLE findings highlight policy silence. Both classifications include drafted suggested policy language for human review.

Yes, you can draft suggested policy language automatically. The monitoring process classifies identified gaps and generates draft updates for human review to realign governance documents with actual AI deployments.