EU Digital Omnibus on AI heads for final approval in June — first amendments to EU AI Act since June 2024 adoption, publication expected in July
The Digital Omnibus on AI — the provisional agreement reached May 7 between the Council, Parliament, and Commission — is on track for formal adoption by the end of June with publication in the Official Journal expected in July. The package is the first set of amendments to the EU AI Act since its June 2024 adoption and reshapes timeline, scope, and obligations for high-risk systems.
The timeline relief is the substantive piece. The Omnibus pushes most high-risk-system obligations into 2027, gives general-purpose-AI providers additional time to align Code-of-Practice signatures, and clarifies the boundary between high-risk and general-purpose obligations. For frontier labs preparing August 2 transparency disclosures, the relief on adjacent high-risk obligations reduces the procedural surface they need to certify simultaneously.
The Brussels signal is calibrated. The amendments do not gut the AI Act — Article 5 prohibitions, transparency obligations, and the GPAI Code of Practice remain. What changes is operational timing and clarification on overlap with existing horizontal legislation (GDPR, DSA, Data Act). For the labs running multi-jurisdiction compliance programs simultaneously — Anthropic with Glasswing partners in Japan and the US, Google with Gemini in Apple Intelligence, OpenAI with the recently-announced Oracle Cloud distribution — the Omnibus is the most concrete EU compliance signal since the original AI Act adoption. The Code of Practice on content marking is the first operational deliverable in that package.
Global Policy Watch — EU AI Act Update: Timeline Relief, Targeted Simplification, and New Prohibitions → · Mind Foundry — AI Regulations around the World - 2026 → · Wilson Sonsini — 2026 Year in Preview: AI Regulatory Developments for Companies to Watch Out For →