Artificial Intelligence: Council and Parliament agree to simplify and streamline rules

The Council of the EU announced on 7 May 2026 that its presidency and European Parliament negotiators had struck a provisional deal on simplifying some of the EU's artificial intelligence rules. The press release was later updated on 18 May 2026 with a letter from the Council presidency to the Parliament. The proposal belongs to the Omnibus VII package in the EU simplification agenda. Marilena Raouna, Cyprus's Deputy Minister for European affairs, is quoted saying the deal lowers recurring administrative costs, brings legal certainty and boosts protection of children.

The Commission had proposed postponing the high-risk AI rules by up to 16 months, extending some SME exemptions to small mid-caps and strengthening the AI Office. Because the high-risk provisions were due to apply from 2 August 2026, the co-legislators gave the file top priority and largely kept the direction of the Commission's text.

The main changes are as follows. A new prohibition covers AI practices that produce intimate or sexual content without consent, as well as child sexual abuse imagery. Fixed new application dates are set for high-risk rules: 2 December 2027 for stand-alone systems and 2 August 2028 for systems embedded in products. Providers must again register in the EU database when they consider a system exempt from high-risk classification, and the strict necessity standard returns for processing special categories of personal data to detect bias. National AI regulatory sandboxes are postponed to 2 August 2027, and the grace period for transparency solutions on AI-generated content shrinks from 6 months to 3, ending 2 December 2026. The machinery regulation is also exempted from direct application of the AI act.

Both institutions must still endorse the agreement, after which come legal and linguistic revision and formal adoption in the coming weeks.