European Commission opens consultation on draft guidance defining 'high‑risk' AI under Article 6

The European Commission on May 19 published draft guidelines on how to classify “high-risk” AI systems under Article 6 of the EU AI Act and opened a targeted consultation running until June 23, 2026. The draft is aimed at providers, deployers and market surveillance authorities that need to decide whether a system falls into the high-risk bucket under the law. The Commission says the text is intended to support “uniform application and effective enforcement” of Article 6. The consultation page says feedback will be used in the final version before adoption by the Commission. ### What exactly is the Commission asking people to comment on? The Commission’s consultation is focused on the clarity of the draft guidelines and the usefulness of the examples included in them. Registered participants receive an anonymous online questionnaire, and the Commission says only responses submitted through that questionnaire will be reflected in the final summary report. The target audience is broad: providers, developers, organizations using AI systems, public authorities, researchers, civil society groups, supervisory bodies and members of the public. (digital-strategy.ec.europa.eu) The consultation opened on May 19, 2026, and closes on June 23, 2026, at 22:00 CET, according to the Commission’s consultation page. The draft is also presented in a “user-friendly format” on the AI Act Single Information Platform, where users can browse summaries and examples by use case. (digital-strategy.ec.europa.eu) ### How does Article 6 decide whether an AI system is high-risk? Article 6 sets out two main routes. The first, in Article 6(1), covers AI systems that are safety components of products — or are themselves products — covered by EU harmonization laws listed in Annex I, where the product must undergo third-party conformity assessment. The second, in Article 6(2), covers AI systems listed in Annex III. (digital-strategy.ec.europa.eu) The Commission’s draft follows that structure. Its published overview says the guidelines are divided into sections matching Article 6, with one section dealing with Annex I product-related systems and another dealing with Annex III use cases. The Commission says the examples in the draft are meant to cover areas and use cases across the law, but are not exhaustive and may be updated over time. (digital-strategy.ec.europa.eu) ### When can an Annex III system avoid being treated as high-risk? Article 6(3) creates an exemption for some Annex III systems. The AI Act Service Desk says an Annex III system is not considered high-risk if it does not pose a significant risk of harm to health, safety or fundamental rights, including because it does not materially influence decision-making outcomes. The text then lists conditions such as performing a narrow procedural task, improving the result of a previously completed human activity, detecting patterns without replacing human assessment, or carrying out a preparatory task. (digital-strategy.ec.europa.eu) The same provision also sets a clear floor. The AI Act text says an Annex III system is always high-risk if it performs profiling of natural persons. Providers that conclude an Annex III system is not high-risk must document that assessment before placing it on the market or putting it into service. ### Who is this guidance meant to help in practice? (ai-act-service-desk.ec.europa.eu) The Commission says the draft is meant to help providers and deployers assess whether their system is high-risk and, if so, which classification route applies. It also says the guidance is for competent market surveillance authorities, which will be involved in enforcement. On the Commission’s policy page, the institution says the text is not legally binding but reflects its interpretation and will guide enforcement. (ai-act-service-desk.ec.europa.eu) The consultation survey adds that the draft deals exclusively with classification, not the full set of downstream compliance duties for high-risk systems. It also says further Commission guidance will follow on other parts of AI Act implementation. ### What dates matter after this consultation closes? The Commission’s policy page says feedback from this consultation will be incorporated before the final guidelines are adopted. (digital-strategy.ec.europa.eu) The same page sets out the revised enforcement timeline following the political agreement on the “AI Omnibus”: rules for certain high-risk areas such as biometrics, critical infrastructure, education, employment, migration, asylum and border control apply from December 2, 2027, while rules for systems integrated into products such as robotics and industrial machinery apply from August 2, 2028. (ec.europa.eu) (digital-strategy.ec.europa.eu)