EU AI Omnibus: Parliament Plenary Set to Vote on AI Act Amendments

The March 25-26 Brussels plenary votes on the Digital Omnibus package including AI Act implementation timeline extensions and the AI nudifier ban.

EU AI Omnibus Gets Plenary Vote: Timeline Extensions and Hard New Prohibitions

The European Parliament's March 25-26, 2026 plenary session in Brussels voted on the Digital Omnibus package — a significant bundle of amendments to existing digital regulations, including changes to the implementation timeline of the EU Artificial Intelligence Act and the introduction of new hard prohibitions on AI systems that generate non-consensual intimate imagery. The package had been approved by committee with 101 votes in favour, 9 against, and 8 abstentions, and the plenary vote was expected to confirm that strong majority.

The most consequential timeline change extends the compliance deadline for high-risk AI systems in sensitive sectors — biometrics, critical infrastructure, education, employment, law enforcement, and border management — from August 2, 2026 to December 2, 2027. This extension reflects the practical reality that many affected organisations, particularly smaller companies and public sector bodies, are not yet in a position to complete the conformity assessments and documentation requirements that high-risk system classification entails. AI industry groups had lobbied intensively for timeline relief, arguing that the original schedule was technically and operationally unrealistic.

The new prohibition on AI nudifier systems — AI applications that generate sexualised or intimate images of real individuals without their consent — was the most publicly visible element of the package. The prohibition was accelerated by the scandal surrounding Grok, Elon Musk's AI assistant on the X platform, which was found to have generated more than 3 million sexual AI images in an eleven-day period at the end of 2025, including 20,000 involving children. The speed with which this provision moved from political demand to law reflects the exceptional political salience of AI-enabled abuse and the EU's determination to use the AI Act as a genuine governance tool rather than a symbolic statement.