EU publishes draft guidance on high-risk AI classification under AI Act

The European Commission has opened a consultation on draft guidance related to classification of high-risk AI systems under the Artificial Intelligence Act.

The document is intended to help providers, deployers, and regulators determine when AI systems fall within the Act’s high-risk category. It includes practical examples showing how classification criteria could apply across different use cases.

Under the AI Act, high-risk systems are subject to stricter obligations because they may affect areas such as health, safety, access to services, employment, education, or fundamental rights.

The guidance is important because classification determines whether organisations face additional compliance requirements, including risk management obligations, technical documentation, human oversight measures, and conformity assessments.

The Commission said the draft aims to support more consistent interpretation of the rules before enforcement obligations begin to apply more broadly across member states.

The consultation is open through the AI Act Single Information Platform until 23 June 2026. Businesses, public authorities, researchers, civil society groups, and individuals can submit comments on the proposed approach.

The process is a part of a wider implementation challenge surrounding the AI Act. Much of the regulation’s practical impact depends not only on the legal text itself, but also on how categories such as ‘high-risk’ are interpreted operationally by regulators and market actors.