EU AI Omnibus Deal Reached: More clarity on AI Act obligations

In the early hours of 7 May 2026, after roughly nine hours of negotiations, the European Parliament, Council, and Commission reached political agreement on the AI Omnibus during the third political trilogue meeting. The deal, struck after the previous trilogue collapsed on 28 April over the question of how the AI Act should interact with sectoral product safety legislation, brings to a close several months of intense negotiations and clears the path for postponement of the AI Act's high-risk obligations before the original 2 August 2026 deadline takes effect.

What is the AI Omnibus?

The AI Omnibus is a legislative proposal published by the European Commission on 19 November 2025, forming part of the broader Digital Simplification Package. Its goal is to amend the EU AI Act to simplify certain obligations and, most importantly for industry, postpone the application of high-risk AI compliance requirements that would otherwise have applied from August 2026. The package was prompted by widespread acknowledgement that the supporting ecosystem (harmonised standards, conformity assessment infrastructure, regulatory guidance, and notified body capacity) would not be in place in time for the original deadline.

High-risk AI deadlines: delays confirmed

The headline element of the deal is the confirmation of postponement dates for the AI Act's high-risk obligations, an outcome that had already been broadly agreed by Council and Parliament before the trilogue process began:

These fixed dates replace the conditional mechanism originally proposed by the Commission, providing greater legal certainty for providers and deployers building their compliance programmes.

What is no longer directly covered by the EU AI Act

The most contested element of the trilogue, and the issue that caused the 28 April collapse, were the contents of Annex I, specifically whether AI systems already covered by EU sectoral product safety legislation should remain under the AI Act's direct scope (Section A) or be governed primarily by their sectoral frameworks (Section B). The Parliament had pushed for a transfer of all Section A products to Section B which the Council initially resisted.

This sectoral architecture compromise is the single most consequential outcome for industrial AI providers operating across the EU.

New prohibitions on non-consensual intimate imagery

The deal introduces new prohibitions in the AI Act's Article 5 on prohibited practices:

These prohibitions, which were not in the Commission's original proposal but were added during the legislative process with strong cross-party support, will apply from 2 February 2027, giving providers a transition period to ensure compliance. The agreed text also clarifies that the prohibitions are without prejudice to remedies available under national laws for individuals to protect their fundamental rights.

Transition periods narrowed

Several transition periods sought by industry were narrowed during the negotiations:

Next steps?

The political agreement reached on 7 May is not yet the final text. A technical meeting is expected before the end of this week to finalise the wording, particularly within the recitals. The Parliament and Council will then need to formally adopt the consolidated text, with publication in the Official Journal of the European Union (OJEU) anticipated within roughly two months. This timeline should allow the postponement of the high-risk AI obligations to take legal effect before the original 2 August 2026 deadline.

techUK International Policy and Trade Programme activities

techUK supports members with their international trade plans and aspirations. We help members to understand market opportunities, tackle market access barriers, and build partnerships in their target market. Visit the programme page here.