The Digital Omnibus delays the AI Act's high-risk obligations to December 2027 — buying industry eighteen months while the general-purpose rules land now
Signed 8 July 2026, the Digital Omnibus on AI moves stand-alone Annex III high-risk systems — recruitment, credit scoring, education, law enforcement, border control — to 2 December 2027, and product-embedded AI to 2 August 2028. The general-purpose model duties were pointedly left on the August 2026 timetable, a split that tells you which obligations the Commission judged ready to enforce.
The delay is a concession to unreadiness, not a softening of intent. The high-risk conformity regime depends on harmonised standards that are not finished; enforcing against them now would mean penalising providers for failing a test that has no answer key. Deferring to December 2027 admits that plainly, and the same document keeps the general-purpose rules on schedule precisely because those obligations are self-contained enough to enforce.
The geopolitics run underneath. China began enforcing its companion-AI rules on 15 July, and the US continues to weigh federal preemption of state AI laws. The EU's move to enforce general-purpose duties on time while relaxing the high-risk calendar is a bet that it can be the jurisdiction that regulates the model layer without strangling the application layer.
For providers, the split calendar is a planning gift and a trap. The gift is eighteen extra months for high-risk product work. The trap is assuming the whole Act slipped — it did not, and a company that reads 'AI Act delayed' and stops its general-purpose compliance work will discover on 2 August that the finable obligations arrived exactly on time.
Technology.org — EU AI Act: what actually applies on 2 August 2026 → · Cubbbix — AI regulation July 2026: EU August deadline, US preemption & 15 countries →