Eligible companies are making progress towards long-sought guidance on high-risk artificial intelligence systems under EU AI law.
After various delays, the European Commission published draft guidelines and launched a public consultation on May 19 aimed at assisting “providers, adopters and other relevant actors in determining whether an AI system falls into a high-risk category.” The three-step guide provides clarity on implementing high-risk requirements and provides examples that “demonstrate how classification should be evaluated in different domains and use cases.”
While the first section of the guide describes the general principles for classifying high-risk AI, there are two sections that detail the approach and specific requirements for the two high-risk categories under section 6 of the Act. Article 6(1) and Annex I apply to AI systems used in the context of product safety, while Article 6(2) and Annex III cover high-risk systems deployed across eight sectors, including biometrics, education, employment and law enforcement.
The committee said the examples provided in the guidance “seek to cover all areas and use cases, but are not considered exhaustive and may be updated over time.”
The comment period for the draft guidelines is open until June 23. It is unclear when the commission will finalize its guidance or whether it will seek further public comment on subsequent drafts.
The IAPP Resource Center offers a series on the operational implications of the AI Act, including a detailed analysis of the obligations for non-providers and providers of high-risk systems, respectively.
hold up
The guidelines were originally scheduled to be submitted on February 2nd, ahead of the AI Act’s provisions on high-risk systems coming into effect on August 2nd.
Following the delay of the original February deadline, a European Commission spokesperson told Euractic that further delays are imminent as the draft guidance is “currently subject to schedule amendments” to ensure it takes into account and incorporates the feedback of as many stakeholders as possible.
The European Commission’s delays played a key role in speeding up the recently finalized Digital Omnibus on AI, which was first introduced in November 2025. Stakeholders, including more than 110 EU-based companies, lobbied for a two-year moratorium on implementation of the high-risk rules, mainly due to the lack of adequate guidance and standards amid a shrinking grace period.
A tentative agreement between EU institutions on AI law reform included an extension of deadlines for compliance with high-risk rules. The effective dates for the rule for standalone high-risk systems are December 2, 2027, and August 2, 2028 for high-risk systems integrated into products.
At a hearing in the European Parliament’s Civil Liberties, Justice and Home Affairs Committee on January 26, Renate Nicolai, the European Commission’s Deputy Director-General for Communications Networks, Content and Technology, said the AI omnibus “gives us a little more time to work on either guidelines, specifications or standards, providing legal certainty for industry and innovators, and allowing us to have a complete system in place.”
