EU AI Act: what was delayed — and what was not
Last updated 1 August 2026
“The EU delayed the AI Act” is the most repeated — and most misleading — compliance headline of 2026. It is half true. The Digital Omnibus on AI (Regulation (EU) 2026/1744, published in the Official Journal on 24 July 2026 and in force) did postpone a major block of the AI Act. But if your exposure is a chatbot, a generative feature or AI-published media, the deadline that applies to you did not move. Here is the clause-level picture, sourced from the amended Article 113.
What was delayed
- High-risk AI systems under Annex III (employment, education, credit, essential services, law enforcement etc. — Article 6(2)): the obligations of Chapter III, Sections 1–3 now apply from 2 December 2027 instead of 2 August 2026.
- High-risk AI systems embedded in regulated products under Annex I (machinery, medical devices, vehicles etc. — Article 6(1)): Chapter III, Sections 1–3 apply from 2 August 2028 instead of 2 August 2027.
- Newly added prohibitions (Article 5(1), points (ba) and (bb), and Article 5(1a) and (1b)) apply from 2 December 2026.
What was not delayed
- Article 50 transparency obligations (Chapter IV): chatbot disclosure, machine-readable marking of synthetic content, emotion-recognition disclosure, deep-fake and AI public-interest text labelling. These apply from 2 August 2026 — the Digital Omnibus left the date untouched. See who must do what now that Article 50 is in force.
- Penalties (Article 99): the fine framework — including up to €15M or 3% of worldwide turnover for transparency breaches under Article 99(4) — applies as part of the general application. See what you actually risk.
- Prohibitions and AI literacy (Chapters I–II): already applicable since 2 February 2025.
- General-purpose AI model rules(Chapter V): applicable since 2 August 2025; the Commission's enforcement powers over GPAI providers switch on with the general application date.
The date table after the Digital Omnibus
- 2 February 2025 — prohibitions (Article 5) and AI literacy (Chapters I–II).
- 2 August 2025 — governance and general-purpose AI model obligations.
- 2 August 2026 — general application date, including Article 50 transparency. Unchanged.
- 2 December 2026 — newly added Article 5 prohibitions (Digital Omnibus).
- 2 December 2027 — Chapter III, Sections 1–3 for Annex III high-risk systems (moved from 2 August 2026).
- 2 August 2028 — Chapter III, Sections 1–3 for Annex I embedded high-risk systems (moved from 2 August 2027).
Why the misunderstanding is expensive
Teams that read “AI Act delayed” and paused all compliance work are now non-compliant on transparency, the part of the Regulation that is cheapest to implement and easiest for regulators and competitors to spot from the outside: a chatbot with no AI notice is visible to anyone who opens it. Transparency breaches carry fines up to €15M or 3% of worldwide turnover (Article 99(4)(g)) — and the obligations reach non-EU providers whenever the output is used in the EU (Article 2).
The fix is fast: run the free 3-minute Article 50 check to see which paragraphs apply, then deploy disclosures with the badge and mark generated media via the C2PA API — or compare plans.
Sources
Not sure which obligations apply to you?
Run the free 3-minute Article 50 check