Two European Parliament committees voted in favor of an omnibus amendment to the EU AI Act, proposing deadline extensions for high-risk systems and a ban on AI-generated nonconsensual sexual imagery. The vote, announced by the European Parliament, passed 101 in favor, 9 against, and 8 abstentions, according to the Parliament’s announcement.
Three compliance deadlines would change under the proposal. Listed high-risk AI systems, those falling under Annex III of the Act, would gain 16 months of additional runway, with their deadline moving to December 2, 2027. AI systems already covered by EU sectoral safety legislation, such as medical devices and machinery, would face a compliance deadline of August 2, 2028. Providers of AI systems generating audio, image, video, or text content would, according to the committee proposal, face a watermarking compliance requirement by November 2, 2026.
The amendment also adds a prohibition that wasn’t in the original AI Act. Insurance Journal’s coverage of the vote directly quotes the draft language: AI systems that generate realistic images “so as to depict sexually explicit activities or the intimate parts of an identifiable natural person” without consent would be banned, with an exception for systems incorporating effective safety measures.
Not everything is delayed. GPAI obligations, prohibited practices already in effect, and compliance requirements that have already passed are unaffected by this proposal.
The committee approval aligns Parliament’s position with EU member states on the nonconsensual image ban, according to Insurance Journal and Reuters reporting, a step toward final adoption. The proposal has not yet been enacted into law.
Compliance teams should identify which deadline category applies to their systems now, the three tracks carry different timelines and different requirements.