Strategic Report  ·  2026-05-24

Draft Commission Guidelines on the Classification of High-Risk AI Systems

Strategic ReportHigh impactEuropean Union
The European Commission published draft guidelines on 19 May 2026 clarifying how to determine whether an AI system qualifies as 'high-risk' under Article 6 of the EU AI Act, and opened a targeted stakeholder consultation running until 23 June 2026. The guidelines set out the Commission's authoritative interpretation of the two classification scenarios defined in the Act — systems listed in Annex III and systems embedded in products covered by specific Union harmonisation legislation — accompanied by a curated set of practical examples spanning health, biometrics, critical infrastructure, employment, education, and law enforcement. Although not legally binding, the document represents the Commission's definitive interpretive position and will guide enforcement by national market surveillance authorities once the high-risk provisions take full effect. The draft is presented in modular, section-by-section form on the AI Act Single Information Platform to facilitate targeted feedback, with a final version to be adopted by the Commission after incorporation of consultation responses.
Any organisation providing or deploying AI systems in the EU must now assess whether its products fall within the high-risk tier; the practical examples in these guidelines are the clearest signal yet of how regulators will draw that line. Legal, compliance, and product teams have until 23 June 2026 to submit feedback that could shape the final, binding text.
Forward the draft guidelines and examples to legal/compliance and product teams; map your EU-facing AI deployments against the classification criteria before the 23 June feedback deadline and consider submitting a formal response.
Sources
European Commission — Shaping Europe's Digital FutureEuropean Commission — Guidelines for Providers and Deployers of AI High-Risk Systems
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →