Cybrilliant Talk to an Expert

Decision Framework

AI Risk Register: What Should Be Tracked?

Track the paths into and out of each AI system, not generic AI risks. A framework for deciding what belongs in the register, at what level, and when to re-rate.

Cybrilliant AI Security SME17 min read

What’s inside

  1. The short version
  2. Track pathways, not categories
  3. Why AI registers go stale
  4. Start with an inventory you can trust
  5. Three levels of entry
  6. What earns a system entry
  7. What to track for each system
  8. Write scenarios, not categories
  9. Use frameworks as a completeness check
  10. Rating and decisions
  11. Triggers that force a re-rating
  12. A worked example: two entries side by side
  13. What to leave out
  14. Indicators worth tracking
  15. Where to start

Sources are listed in the full decision framework.

Members-only decision framework

Get the full decision framework.

Enter your details and we will email you the PDF of “AI Risk Register: What Should Be Tracked?”.

Related Insights

Keep reading