Serious Incidents. Without Undue Delay.
EU AI Act Art. 73 mandates serious incident reporting within strict deadlines. OmegaOS automates classification, timeline tracking, and authority notification — so you never miss a deadline.
What Art. 73 Requires
Providers and deployers of high-risk AI systems must report serious incidents to national competent authorities. The obligations are precise, the deadlines non-negotiable.
| Obligation | Deadline | What OmegaOS does |
|---|---|---|
| Report serious incidents to national authority | Without undue delay (initial); complete report within applicable period | Automated initial report generation, deadline countdown, authority notification tracking |
| Classify incident severity | Immediately upon detection | Three-tier classification: Critical (2 days), Major (10 days), Standard (15 days) |
| Document corrective actions | Within reporting period | Structured corrective action tracking with status workflow |
| Notify affected parties | Without undue delay | Impact assessment with affected party enumeration |
Three Severity Tiers
Each incident is classified by severity. The tier determines the investigation window, the reporting deadline, and the escalation protocol.
Tier 1 — Critical (2 days)
Immediate threat to health, safety, or fundamental rights. Automated escalation. Authority notification within 48 hours.
Tier 2 — Major (10 days)
Significant impact on affected persons. Extended investigation window. Full report with corrective actions.
Tier 3 — Standard (15 days)
Material impact requiring documentation. Standard investigation. Complete report with root cause analysis.
Incident Lifecycle
From detection to closure, every step is tracked, timestamped, and auditable.
Post-market monitoring flags anomaly
Automatic severity classification (Tier 1 / 2 / 3)
Generated within hours, structured for authority submission
Track corrective actions, impact assessment, root cause
Record notification timestamp, track acknowledgment
Full investigation report with all evidence
Deadline Intelligence
Every incident has a countdown. OmegaOS tracks days remaining, warns when deadlines approach, and flags overdue incidents. The dashboard shows real-time status across all open incidents — no spreadsheet required.
API Endpoints
Programmatic access to the full incident lifecycle.
POST /api/v1/tenants/{id}/incidents — Create incident
GET /api/v1/tenants/{id}/incidents — List incidents
GET /api/v1/tenants/{id}/incidents/overdue — Overdue incidents
POST /api/v1/.../incidents/{id}/notify-authority — Record notification
GET /api/v1/.../incidents/{id}/report — Get report (JSON/Markdown) Connection to Art. 72
Art. 72 (post-market monitoring) detects. Art. 73 (incident reporting) reacts. OmegaOS connects both — drift alerts from the monitoring pipeline can trigger incident classification automatically.
When KL divergence exceeds policy thresholds, the monitoring system generates a signal. That signal can be routed directly into the incident pipeline, pre-classified by severity, with the clock already running.
Disclaimer
This page describes technical capabilities. It does not constitute legal advice and does not imply certification or formal compliance. Obligations depend on role (provider, deployer, importer), use case, and jurisdiction. Consult qualified legal counsel.
Compliance is a deadline, not a destination.
Discuss how OmegaOS integrates incident reporting into your compliance workflow.
Book a Pilot Call