The Digital Omnibus took a decisive step on 2 June 2026, when the Internal Market and Civil Liberties committees backed the provisional deal on simplifying EU artificial intelligence rules. The two committees adopted the agreement by 93 votes in favour, four against and 15 abstentions. Negotiators from Parliament and the Cyprus Presidency of the Council had concluded the compromise in the early hours of 7 May.

The package forms part of the Commission’s wider push to cut red tape in digital law. Its most urgent element concerns the AI Act, whose obligations for high-risk AI systems and models were due to bite from 2 August. Lawmakers wanted to postpone those duties until the technical norms and standards that companies need are available.

That timing explains the pressure on the calendar. Adoption before the August deadline was a priority, because companies would otherwise face legal obligations without agreed standards to follow. The Digital Omnibus therefore combines a legal fix with a political message that Europe wants workable rules rather than paper obligations.

Supporters argue that the Digital Omnibus delay protects innovation. Developers of medical devices, hiring tools and credit scoring systems face the problem that they cannot certify products against standards that do not yet exist. A pause gives standardisation bodies time to finish their work and gives smaller firms time to prepare for audits, documentation and testing duties.

Critics see risks. Postponing high-risk obligations could leave people exposed to unreliable systems for longer. The 15 abstentions in committee show that some members remain uneasy, even though the margin of 93 to four leaves little doubt about the majority view.

The agreement also reflects a political bargain with a protective element. Reporting on the deal says it bans so-called digital nudifier apps, which generate fake intimate images. That addition lets lawmakers present the Digital Omnibus as more than deregulation.

The standards question deserves a closer look. European standardisation bodies have been drafting technical norms for high-risk AI systems, and several deliverables slipped past their original deadlines. Without them, notified bodies cannot assess conformity in a consistent way, and companies cannot show regulators that they comply.

Procedure still matters. After the committee vote, a plenary vote was scheduled for June, and the Council must also formally approve the text. Readers should check the Official Journal for the final outcome, because the committee report does not confirm either step of that process.

For companies, the practical message is to keep preparing. Even with a delay, compliance teams should map which of their systems count as high risk and follow the work of European standardisation bodies. The Digital Omnibus buys time but does not remove the obligations, and firms that wait for the final text risk losing months they cannot recover.

For Brussels, the file is a test case for the Commission’s simplification agenda. If the Digital Omnibus delivers legal certainty before the August deadline, the Commission will point to it as proof that simplification works. If it slips, critics will say that the EU keeps writing rules faster than it can make them usable.