At 4:30 a.m. on 7 May 2026, the European Parliament and the Council of the EU reached a provisional agreement on the Digital Omnibus on AI — the first significant set of amendments to the EU AI Act since its adoption in June 2024. The deal recalibrates timelines and introduces targeted simplifications, while adding a new prohibition that had broad political support across institutions.
The headline change for compliance teams is the proposed deadline extension. If formally adopted in the agreed form, standalone high-risk AI systems classified under Annex III would move from 2 August 2026 to 2 December 2027. High-risk systems embedded in regulated products under Annex I would move to 2 August 2028. Where sector-specific legislation — including the Machinery Regulation and medical devices rules — already imposes AI-related requirements, the deal would introduce a mechanism to limit the AI Act's application through implementing acts, with the Machinery Regulation receiving a full exemption.
The agreement would also introduce the EU's first explicit prohibition on nudifier applications — AI systems designed to generate non-consensual sexually explicit or intimate content, including child sexual abuse material. Providers would be required to comply with the new prohibition by 2 December 2026, giving roughly six months from formal adoption. The move follows widespread concern over the proliferation of deepfake intimate imagery, with the European Parliament Research Service citing approximately 8 million deepfakes online in 2025.
Formal adoption is expected before 2 August 2026, the date on which the current high-risk rules would otherwise have become applicable. Upon adoption, the amendments will be published in the Official Journal and enter into force three days later.
Acompli perspective: This is the update to the Council mandate we covered in March — now with Parliament's input baked in, but still subject to formal adoption and Official Journal publication. The proposed extended deadlines do not reduce the compliance workload; they redistribute it. Organisations should use any additional runway to strengthen their impact assessments for AI systems, refine risk classification documentation, and build the internal governance structures that the Act demands. The nudifier prohibition, if adopted, carries a much shorter timeline and requires immediate attention from any organisation deploying generative AI capable of producing imagery.