On 24 June 2026, the Council of the European Union and the European Parliament adopted the Digital Omnibus on AI, a legislative act amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 to simplify harmonised rules on artificial intelligence. The act aims to reduce administrative burdens for businesses while maintaining safety and fundamental rights protections, impacting AI developers, deployers, and national market surveillance authorities across the EU.
The document, published as an 'I/A' item note, was approved by the Council on 24 June 2026 following the Parliament's adoption. It introduces targeted amendments to the AI Act, the Aviation Safety Regulation, and the Machinery Regulation to streamline compliance for high-risk AI systems.
Belgium supported the compromise text but attached conditions and regrets. It backed the text contingent on an amendment banning AI systems that generate non-consensual sexually explicit content. However, Belgium regretted the postponed grace period for labelling AI-generated content, citing technical challenges that risk individual protection and deepfake transparency. It also welcomed clarifications on bias detection but regretted that 'strict necessity' was not further clarified and that synthetic data alternatives were not further encouraged.
The Commission raised concerns about implementation deadlines. It must adopt delegated acts by 2 August 2027 specifying conditions for limiting certain AI Act requirements for high-risk AI systems under Article 6(1) AI Act, where sectoral legislation ensures equivalent protection. The Commission warned that this deadline risks turning the delegated-act mechanism into a one-time exercise, limiting its ability to address future overlaps. Additionally, the Commission must develop guidelines by 1 August 2027 on the interplay between high-risk AI requirements and sectoral legislation. The Commission cautioned that this deadline, a full year before high-risk requirements apply to Article 6(1) systems on 2 August 2028, may jeopardise reliable support tools for industry and authorities.
AI developers face reduced compliance costs due to simplified rules, but may need to adapt to new delegated acts and guidelines by 2027-2028. National authorities gain clearer oversight but must prepare for the new framework. Consumers benefit from continued protections, though delayed deepfake labelling may temporarily reduce transparency. The Commission's tight deadlines may limit its ability to refine rules, potentially creating uncertainty for industry.