AI Risk Manager — Compare Workflow¶
Description: Evaluate multiple approaches or versions
When to Use¶
Use the compare workflow when you need to evaluate multiple approaches or versions.
Input Requirements¶
- NIST AI RMF profiles and playbooks
- AI system threat models and attack surface analyses
- Risk register entries and historical incident data
- Regulatory risk requirements (EU AI Act, sector-specific regulations)
Process¶
- Initialize — Set up the compare context for AI Risk Manager
- Execute — Perform the compare operation following AI Risk Manager's style
- Validate — Check output against quality gates
- Handoff — Deliver results to downstream personas
Output¶
- AI risk registers with likelihood, impact, and mitigation status
- Risk heat maps showing portfolio-level AI risk exposure
- NIST AI RMF function mapping reports (Govern, Map, Measure, Manage)
- Continuous monitoring dashboards with risk trend analytics
Quality Gates¶
- Risk assessments must cover all four NIST AI RMF functions
- High-risk AI systems require continuous monitoring, not just initial assessment
- Risk appetite and tolerance levels must be defined by governance authority
- Emerging risks must be captured within one assessment cycle of identification