The European Commission published draft guidelines clarifying which artificial intelligence systems fall under the EU AI Act’s high-risk category, according to a notice on its digital strategy site. The non-binding text is intended to help providers and deployers assess whether their systems trigger the Act’s most demanding obligations.
The draft matters because the AI Act imposes compliance requirements on high-risk systems that affect health, safety or fundamental rights. Without clear classification, companies face uncertainty about which rules apply.
The guidelines were developed after a public consultation and input from member states through the AI Board. They present examples covering areas such as biometrics, critical infrastructure, education and employment, though the Commission stresses the list is not exhaustive and may be updated.
The document notes that, following the political agreement on the “AI Omnibus”, enforcement deadlines have been revised. For systems used in critical domains—including biometrics, critical infrastructure, education, employment, migration, asylum and border control—the rules will apply from 2 December 2027. For AI integrated into products such as robotics and industrial machinery, the effective date is 2 August 2028.
A targeted stakeholder consultation on the draft closed on 23 July 2026. The Commission said it will incorporate feedback before adopting a final version, but it has not announced when that adoption will occur. The draft guidelines are available on the AI Act Single Information platform.
The Commission notes that the text is not legally binding but “reflects the Commission’s interpretation and will guide enforcement”. No independent analysis of the guidelines was provided.