AI Incident Escalation and Notification

Detect, triage and escalate serious AI incidents and relevant risk conditions within required timelines.

Give staff a clear way to report suspected serious AI incidents and relevant risk conditions. Assign a response lead and record when the organization became aware of the issue. Determine the actor, system and applicable notification route, then contact the required parties within the relevant timescale. For deployers, distinguish provider-first serious-incident escalation from the risk-and-suspension route. Use the High-Risk AI Serious Incident Reporting control for the high-risk provider reporting procedure. Real-world testing also has immediate mitigation, suspension or termination duties; its incident reporting follows the applicable statutory route. Protect evidence and track follow-up rather than waiting for a complete investigation before escalating.

On incident or relevant risk detection; immediately or within the applicable legal deadline; periodic exercises set by the organization.

Incident intake record with awareness time Actor-specific escalation matrix and current contacts Classification, notification and suspension decisions Transmission records and response actions Exercises, findings and corrective follow-up

Walk through a realistic incident scenario with the response team. Check that they can identify the required route and contact sequence. Review an actual incident where available, including the recorded awareness time and response evidence.