Draft Commission guidelines on classifying high-risk AI systems published for consultation

On 19 May 2026 the Commission published draft guidelines on the classification of high-risk AI systems under Article 6 of the AI Act, with practical examples of systems that should and should not be classified as high-risk — covering both safety components of regulated products (Annex I) and the Annex III use cases. The draft was put to a targeted stakeholder consultation, with final guidelines to follow.

High-risk classification drives the heaviest obligations in the Act. The draft shows the Commission's intended reading before the rules bite under the revised timetable — early enough to test your classifications and to influence the final text.

Providers and deployers of AI systems near the Annex III boundary, and market-surveillance-facing compliance teams preparing classification files.

Re-screen your Annex III and safety-component classifications against the draft's examples, flag systems whose status would change, and consider responding to the consultation while the draft is open.