The AI Act introduces incident reporting for serious incidents involving AI systems. If your AI system causes harm, is manipulated, or fails in a way that affects people, you may need to report it. Here's what counts, how to report, and how it connects to NIS2.
Not every AI glitch is reportable. A chatbot giving a wrong restaurant recommendation isn't. A hiring AI discriminating against a protected group is.
Serious incidents involving high-risk AI systems must be reported to the relevant market surveillance authority. The specific authority depends on the AI system's domain — employment, healthcare, finance. The AI Act coordinates with national authorities, not a single central body.
If your AI system is also part of your NIS2 scope, an AI incident may also be a NIS2 incident. You'd report to both: the AI Act authority and your national CSIRT. The evidence requirements stack.
If you deploy AI in an NIS2-regulated sector, incident reporting obligations can double. One report to the AI Act authority, one to the CSIRT. Both need evidence.
Aethel Guard is a Netherlands-based security architecture firm serving EU companies of all sizes. Your architect, not your vendor.