AI governance checklist
Use this checklist to create a minimal governance record before an AI system enters production, procurement, or customer-facing use.
Ownership
Name the system owner, model owner, data owner, risk reviewer, and approval path.
Risk tiering
Classify impact, affected users, autonomy level, reversibility, and evidence required before launch.
Monitoring
Define drift, safety, fairness, privacy, and incident triggers with review cadence.
- Keep model purpose and prohibited use cases visible.
- Record datasets, evaluations, limitations, and known residual risk.
- Make rollback and human escalation paths explicit.