Regulation (EU) 2026/1744, the Digital Omnibus on AI, entered into force on Monday — three days after its publication in the Official Journal — and with it the AI Act's most contested deadline stopped existing.

What moved, and how far

Obligations for stand-alone high-risk systems under Annex III — employment, education, biometrics, essential services, migration — shift from August 2, 2026 to December 2, 2027. AI embedded in regulated products under EU harmonisation law, such as machinery, toys and lifts, goes to August 2, 2028. Legacy systems run by public authorities stay where they were, on 2030.

What did not move

This is where most of the coverage goes wrong. Obligations for general-purpose AI models are untouched, and so is the Commission's enforcement power over them: from August 2, 2026 — next Sunday — it can fine providers up to 3% of global turnover or €15 million. The rollback bought time for deployers of high-risk systems, not for model developers.

In force is not the same as applies

Entry into force means the amending regulation is now law. It does not mean every provision is operative today; each applies on its own staggered date. Headlines reading "EU AI Act rules take effect" describe the opposite of what happened.

The rest of the package

The Omnibus also extends simplifications for small mid-caps, creates an EU-level regulatory sandbox, downgrades mandatory AI literacy to non-binding encouragement, widens the AI Office's oversight of general-purpose models, and adds fresh prohibitions on AI systems generating non-consensual intimate imagery and child sexual abuse material.

The decision was old; the effect is new

The political agreement was reached on May 7, the European Parliament voted on June 16 and the Council on June 29. Those were the news events at the time. What changed this week is legal, not political: as of today the August 2 high-risk deadline is off the books.