Incident Management Procedure
Procedure for managing and reporting serious incidents per Article 73 including detection, investigation, notification to authorities, and corrective actions.
40 min
Read Time
Level
STD-AI-013: Incident Management Standard
[To be filled]
[To be filled]
Purpose
This procedure establishes the step-by-step process for detecting, classifying, responding to, investigating, and reporting incidents related to high-risk AI systems in compliance with EU AI Act Article 73. The procedure ensures serious incidents are identified promptly, reported to competent authorities within required timelines, and properly investigated to prevent recurrence and improve AI system safety.
Applies To
- All high-risk AI systems classified per AI System Classification Standard (STD-AI-001)
- All AI systems that could cause serious incidents as defined in Article 3(49)
- Providers and deployers of high-risk AI systems
- All incidents and malfunctions affecting AI systems
- Serious incidents requiring regulatory reporting per Article 73
Does Not Apply To
- Minimal-risk AI systems (unless incident causes serious harm)
- Limited-risk AI systems (unless incident causes serious harm)
- Third-party AI system incidents (covered by vendor escalation in AI Vendor Management Procedure (PROC-AI-VENDOR-001))
INCIDENT DETECTION & CLASSIFICATION
IMMEDIATE RESPONSE & NOTIFICATION
INVESTIGATION & ROOT CAUSE ANALYSIS
REGULATORY REPORTING
CORRECTIVE ACTIONS & CONTINUOUS IMPROVEMENT
13
Procedure Steps
8
Roles Defined