The European Commission has published draft guidelines on the classification of high-risk AI systems under the EU AI Act and opened a targeted stakeholder consultation, which will remain open until 23 June 2026.

The draft guidelines are intended to assist providers and deployers in assessing whether an AI system should be classified as high-risk under the AI Act. According to the Commission, the document reflects feedback collected through consultations with stakeholders, as well as input provided by Member States through the AI Board.

The draft guidelines are organized into three sections: the first introduces the general principles for determining whether an AI system should be considered high-risk and outlines the two categories set out in Article 6 of the EU AI Act, which are examined in detail in the following two sections. The first category concerns AI systems that constitute safety components of products, or that are themselves products, subject to EU harmonization legislation on product safety; the second category concerns autonomous AI systems used in specific sectors listed in Annex III of the AI Act, including biometrics, critical infrastructure, education, employment, migration, asylum and border control, access to essential services and law enforcement.

According to the Commission, the draft guidelines are intended to provide practical assistance through the inclusion of examples covering different sectors and use cases. Annex III of the AI Act lists the general areas that are particularly exposed to risks associated with high-risk AI systems. This list of areas includes, by way of example but not limited to, (i) AI systems intended to be used in the field of biometrics; (ii) AI systems intended to be used in employment, worker’s management and access to self-employment; (iii) AI systems intended to be used to evaluate the creditworthiness of natural persons or establish their credit score, with the exception of AI systems used for the purpose of detecting financial fraud.

The Commission specifies that such examples are not exhaustive and may be updated over time. The draft guidelines further clarify that the inclusion of a use case does not constitute an assessment of the lawfulness of a specific AI system under other applicable legal frameworks.

The Commission also states that the draft guidelines are not legally binding and reflect the Commission’s interpretation of the AI Act.

The final version of the guidelines will be adopted following the conclusion of the ongoing stakeholder consultation.

The publication of the draft guidelines follows the revised implementation timeline introduced in the context of the political agreement on the AI Omnibus package. Under the updated timeline, rules applicable to certain standalone high-risk AI systems will apply from 2 December 2027, while rules applicable to high-risk AI systems integrated into products, including robotics and industrial machinery, will apply from 2 August 2028.

According to the Commission, additional guidance aimed at facilitating compliance with obligations applicable to high-risk AI systems is expected to complement the current draft guidelines over time.