Handle an AI incident
When an AI system misbehaves in production, Governance Fabric stitches together policy retrieval, risk re-assessment, and governance re-scoring against the affected initiative in one flow.
How it works
- 1Locate the affected initiative below and open its detail page.
- 2Query Policy → your incident-response policy from Secure-RAG or the FAQ corpus.
- 3Click Re-assess (with the red-team toggle on) to re-run ARA against the incident type.
- 4The initiative's Trust Score, CALIBRE band, and Decision Pack update automatically to reflect the post-incident posture.
- 5Optionally raise the initiative's lifecycle stage back to Monitor or Design pending remediation.
Pick an initiative
| Initiative | Band | Verdict | |
|---|---|---|---|
Autonomous personal-loan approval (Australia) SAMPLE-01 SAMPLE |
Critical | Escalate to risk committee | Open initiative |
AI-assisted AML transaction monitoring SAMPLE-02 SAMPLE |
High | Approve with conditions | Open initiative |
HR policy RAG chatbot (global workforce) SAMPLE-03 SAMPLE |
Moderate | Approve | Open initiative |
Autonomous SOC response agent SAMPLE-04 SAMPLE |
Critical | Do not approve | Open initiative |
Radiology AI triage assistant (hospital) SAMPLE-05 SAMPLE |
High | Approve with conditions | Open initiative |