AI Digital Omnibus: political agreement to simplify the AI Act and strengthen safeguards

The European Union has taken a further step towards shaping its regulatory framework for artificial intelligence, with a clear focus on current issues and practical implications for organisations. In recent days, the European Parliament and the Council have reached a political agreement on the so-called “AI Digital Omnibus”, as set out in the Commission’s press release of 7 May 2026, which aims to simplify the application of the Artificial Intelligence Regulation (“AIR”), clarify deadlines and reduce certain compliance burdens, whilst maintaining levels of protection for fundamental rights and security.

This political agreement — which still needs to be formally endorsed by the European Parliament and the Council — does not alter the structural principles of the AI Regulation, but it does adjust the timeline and design of certain key obligations. In particular, it establishes a more phased implementation timetable for high-risk AI systems: the rules for systems used in areas such as biometrics, critical infrastructure, education, employment, migration, asylum and border control would apply from 2 December 2027, whilst the obligations for systems integrated into products subject to sector-specific safety regulations (for example, lifts or toys) will begin to apply on 2 August 2028.

From a business perspective, the message is twofold. On the one hand, the new timetable provides additional leeway to adapt high-risk AI systems, review inventories of use cases, refine governance models and coordinate the implementation of the AI Act with other sector-specific obligations. On the other hand, the strengthened ban on certain uses (such as ‘nudification’ applications) and the clarification of obligations for providers and users point to increased regulatory expectations and future oversight, making it advisable to anticipate impact assessments and compliance reviews in the field of AI.

Procedurally, the agreement reached on 7 May 2026 is strictly political in nature and must still be formally adopted by the European Parliament and the Council before taking full legal effect. Once this process is complete, the Digital Omnibus Regulation on AI will be published in the Official Journal of the European Union, and this publication is expected to take place before 2 August 2026, so that the amendments come into force prior to the entry into force of certain original obligations under the AI Regulation.

In this context, the “AI Digital Omnibus” represents an opportunity, but also serves as a wake-up call for organisations that use or develop AI systems. Regulatory simplification and the adjustment of deadlines do not imply a relaxation of requirements, but rather a clearer framework for structuring compliance projects, which makes it advisable to initiate — or update — compliance assessments and the review of AI models as soon as possible, particularly in sectors involving high-risk applications.