← AI Governance Readiness Check
AI Risk Register Template
A useful AI risk register tracks an actual scenario, owner, likelihood, impact, controls, residual risk, action, due date, and review date.
Suggested fields
Risk ID · AI system/use case · risk scenario · category · affected stakeholder/process · likelihood · impact · inherent rating · current controls · control evidence · residual rating · action · owner · due date · status · next review.
Common risk categories
- Data leakage, privacy, retention, or unauthorized training use
- Security vulnerabilities, prompt injection, excessive tool permissions
- Hallucination, inaccuracy, model drift, or unreliable outputs
- Bias, unfairness, accessibility, or harmful downstream impact
- Insufficient human oversight or unclear accountability
- Vendor dependency, model changes, outages, or weak exit rights
- Copyright, confidentiality, contractual, or regulatory exposure
- Customer disclosure or transparency failures
- Operational errors, workflow breakage, or automation without exception handling
- Financial waste where AI cost exceeds measurable benefit
- Reputational harm from inappropriate use or public failures
- Evidence gaps that prevent reconstruction of why a decision was made