EU clarifies high-risk AI rules as deadlines move

The European Commission has published draft guidelines on classifying high-risk AI systems under the EU AI Act, with feedback open between the 19 May 2026 and 23 June 2026. The guidance follows planned delays to key implementation dates, including moving the deadline of 2 August 2026 to 2 December 2027 for Article 6(2) Annex III use cases and the deadline of 2 August 2027 to 2 August 2028 for Article 6(1) Annex I safety component cases.

The draft guidance focuses on when an AI system should be treated as high-risk. It sets out general principles, rules for high-risk AI in regulated products, and high-risk use cases in sensitive areas. Classification turns on whether a system meets the legal definition of AI, its intended purpose, and whether it is part of a regulated product or falls within Annex III categories. Providers carry the primary assessment burden, while deployers, importers and distributors can assume obligations if they modify or repurpose a system.

The guidance also examines the Article 6(3) exemption for Annex III systems that do not pose a significant risk of harm, including where they perform narrow procedural tasks or improve previously completed human activity without profiling. UK organisations are advised to map AI tools and EU operations because the Act can apply where an AI system is placed on the EU market, made available for use in the EU, or produces outputs used there.