Incident response
Incident Response is the workspace log of AI-related events: what happened, severity, status, and optional links to a use case or framework program.
Open Governance → Incident Response in the sidebar.
This is not the same as Pending approvals. An incident can have its own approval fields; those are metadata on the incident, not the approvals queue.
What you see
Chips: Total, Open, Investigating, Mitigated, Closed.
Filter by status, or All. Search in the description or the incident ID (for example values starting with INC-).
The table typically shows Incident ID, AI project (or framework name if no project is set), type, severity, status, occurred date, approved by, and a ⋯ menu.
Create or edit
Add new incident — Admin and Editor. If your plan does not allow more records, you are asked to upgrade.
⋯ menu: Edit for people who can open the menu; Delete for Admin and Editor. Delete hides the row from the list.
The form covers:
- Incident information — link to an AI use case or framework program, type, severity, status, occurred and detected dates, optional model version, harm categories, affected persons, description (required)
- Impact & response — how it relates to AI, immediate mitigations, planned actions
- Approval (optional) — approval status, approved by, date, notes, interim report
Pick reporter and approver from workspace members where those fields appear.
Severity: Minor, Moderate, Serious, Very serious, Critical.
Status: Open → Investigating → Mitigated → Closed.