Privacy incident- Unauthorized access, accidental sharing, breach.

Safety incident- Harmful or unsafe AI output.

Fairness incident- Subgroup harm, biased errors, discriminatory pattern.

Governance incident- Undocumented model changes, missing logs.

1. Report - I provide a clear reporting channel (form/email).
   
   

2. Contain - I limit exposure (disable features, restrict access).
   
   

3. Investigate - I review logs, versions, and user impact.
   
   

4. Notify - I communicate what happened and what I’m doing.
   
   

5. Remedy and prevent recurrence - I fix root causes and update controls.

Complaint pathway: Submit concern → receive acknowledgement → investigation outcome.

Escalation: if unresolved, escalate to the governance owner/privacy lead.

Timelines: I acknowledge within X days, provide an update within Y days, and document outcomes.

“If you feel unsafe or in crisis, contact local emergency services or crisis lines.”