On 7 May 2026, the European Parliament and the Council of the EU reached a political agreement on the AI Omnibus. The AI Omnibus is a legislative package amending Regulation (EU) 2024/1689, the Artificial Intelligence Act. The Commission proposed the package on 19 November 2025. The political agreement is a provisional deal; formal adoption by both Parliament and Council is required before the amendments take effect.
The agreement amends Article 6 and Annexes I and III of the AI Act, which define high-risk AI systems. It inserts a new prohibition in Article 5 against AI practices generating non-consensual intimate or sexual content. Compliance deadlines for high-risk AI systems shift on two tracks. Systems in biometrics, critical infrastructure, education, employment, law enforcement, and border management must comply by 2 December 2027. Systems embedded as safety components in products subject to EU sectoral safety legislation must comply by 2 August 2028. Transparency obligations for AI-generated content move to a deadline of 2 December 2026, shortened from the original six-month grace period. Small mid-cap companies gain access to the SME exemptions in Article 86.
General-purpose AI model providers, high-risk AI system deployers, and synthetic content platforms must update their compliance timelines. The high-risk system deadline moves to 2 December 2027 or 2 August 2028 depending on product category. Providers of synthetic content face an earlier transparency deadline of 2 December 2026. Banks, insurers, healthcare providers, and employers using AI in consequential individual decisions should treat 2 December 2027 as their target compliance date for high-risk system obligations.
The agreement clarifies the AI Office's supervisory competence over general-purpose AI models. National competent authorities retain jurisdiction in specified contexts: law enforcement, border management, judicial authorities, and financial institutions, where the same provider develops both the model and the deployed AI system. AI regulatory sandboxes at member-state level are postponed until 2 August 2027. Formal votes in Parliament and Council are required before the amendments take effect; their timing has not been announced.
Licentium advises AI developers, model providers, and deployers on EU AI Act obligations. Our partner network covers Brussels regulatory counsel and member-state advisers across the EU; contact us to discuss compliance planning. Work we undertake includes high-risk AI system classification, conformity assessment planning, general-purpose AI model documentation, EU AI Act compliance roadmap development, and transparency disclosure design for AI-generated content platforms.