Nicola FabianoAnnex I of the AI Act: the less-scrutinized frontier of high-risk classification.
Not a minor variant of Annex III — the structural point where the AI Act grafts onto EU product safety, conformity assessment, and CE marking.
The Article 6(3) derogation does NOT apply here. And the post-Omnibus narrowing of "safety component" changes the perimeter.
New analysis on my blog NicFab.eu:
🔗 https://www.nicfab.eu/en/posts/annex-i-high-risk-classification/
Annex I of the AI Act: The Less Scrutinised Frontier of High-Risk Classification
The Draft Commission Guidelines of 19 May 2026 and the Omnibus agreement of 7 May 2026 bring renewed attention to an under-examined zone of the AI Act: AI systems embedded as safety components of regulated products. An analysis of Article 6(1), the structure of Annex I, the grafting onto sectoral conformity assessment regimes, and the operational implications for providers and notified bodies.
IECstandards
EURACTIV Technology