The European Parliament approved a simplification proposal for the Artificial Intelligence Act with 569 votes in favor. The measure, described as a digital omnibus, sets fixed application dates intended to give businesses legal certainty and allow time for implementation guidance to be prepared.
Under the revised timeline, requirements for high-risk Artificial Intelligence systems involving biometrics, critical infrastructure, and essential services are proposed to take effect on December 2, 2027, while Artificial Intelligence systems covered by specific EU sectoral safety legislation have a proposed deadline of August 2, 2028. Providers of Artificial Intelligence-generated content must comply with watermarking rules for audio, image, video, and text by November 2, 2026.
The updated position also introduces a ban on nudifier Artificial Intelligence systems that create non-consensual sexually explicit imagery of real people, while exempting systems with effective safety safeguards. Members of the European Parliament also extended support measures to small Mid-cap Enterprises and sought to reduce regulatory overlap by making Artificial Intelligence Act obligations less stringent for products already governed by sectoral laws such as medical devices and toys.
The legislation now moves into negotiations with the Council to determine its final form. The change formally extends the compliance timeline after EU regulators missed internal target dates for industry guidance, giving businesses a more realistic window to adjust. The proposal is part of a broader Digital Omnibus package introduced by the European Commission in late 2025 to harmonize data protection and digital business laws.
