liveBook · Manning

Incident reporting obligation

As explained in AI Governance

An incident reporting obligation is a legal duty to notify designated authorities about a qualifying incident within a specified period.

A reasonable likelihood of a link between an AI system and an incident starts the reporting clock; confirmed root cause is not required. Deployer and provider duties are distinct, and corrective action runs alongside reporting. section 7.5.2

Apply the reporting path:

  • When a deployer spots a qualifying pattern, it must immediately notify the provider, importer or distributor, and relevant market surveillance authority. Draft guidance interprets “immediately” as within 24 hours. section 7.5.2
  • The provider carries the formal reporting obligation and corrective action duty. A GPAI model with systemic risk reports to the AI Office instead of the national market surveillance authority. section 7.5.2

Triage a recurring failure

ClaimAssist wrongfully denied favorable payment-plan terms to dozens of applicants with a chronic condition. The pattern persisted for months, affecting people who did not challenge the terms; it likely crosses the serious-incident threshold on grounds including indirect health harm. section 7.5.2

Figure 7.14 EU AI Act incident reporting tiers. Clock starts at reasonable likelihood of a causal link, not confirmed root cause.

Use the reporting-tier figure to review when the clock starts and how incident reports are routed. section 7.5.2

How AI Governance covers it

How deep it goes
Distinguishes the reasonable-likelihood trigger, reporting roles, authority paths, and a worked harm-pattern threshold assessment.
Examples
Assesses ClaimAssist wrongful payment-plan denials affecting applicants with a chronic condition over several months.
What you can do after
Diagnose whether a recurring harm pattern likely crosses the serious-incident threshold and route notifications to designated recipients.
How it connects
Connects serious-incident reporting to the EU AI Act, market surveillance, providers, deployers, and GPAI systemic-risk oversight.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Governance. The section links point to the parts of the book it draws on.