EU Parliament adopts Digital Omnibus for AI negotiating position
The European Parliament adopted its AI Digital Omnibus negotiating position, proposing later deadlines for high-risk rules and AI-content watermarking requirements.
Summary
The Privacy Newsfeed, a free news digest aimed at data protection officers, carries a short summary reporting that the European Parliament has settled its bargaining stance on the AI part of the Digital Omnibus package. The tally was 569 votes for, 45 votes against and 23 abstentions. The item draws on legal analysis by the firm Pinsent Masons and points readers to the full story elsewhere.
According to the summary, the Parliament's position would push back the start of some high-risk AI system rules under the Artificial Intelligence Act until 2 December 2027, so that relevant guidance and standards can be ready beforehand. Other rules covering AI systems that are subject to sectoral legislation would be put off until 2 August 2028. Providers of AI would receive until 2 November 2026 before obligations to watermark AI-created audio, images, video or text start to apply.
The position also contains a proposal that has no counterpart in the dates above. Members of the Parliament also want to add a further banned practice, covering AI systems that produce sexual and intimate content without the consent of the people depicted.
As for next steps, the summary explains that the vote clears the way for trilogue negotiations to begin involving the Parliament and EU member governments in the Council. The page also carries a training advertisement for a certificate in AI governance and a general description of the newsfeed, neither of which concerns the vote itself.