The EU AI Act applies in stages. Prohibitions and AI literacy apply from 2 February 2025, general-purpose AI rules from 2 August 2025, and most of the rest from 2 August 2026. After the AI Omnibus, high-risk rules for Annex III uses move to 2 December 2027 and product-embedded systems to 2 August 2028.
The staged calendar
The Act entered into force on 1 August 2024, but nothing applied on day one. The European Commission's policy page and the Act's implementation timeline give this sequence:
- 2 February 2025: prohibited practices and AI literacy obligations.
- 2 August 2025: governance rules and obligations for providers of general-purpose AI models.
- 2 August 2026: general applicability, with exceptions. The AI Office and Member State enforcement begin, and the transparency rules take effect.
- 2 December 2026: a new prohibition on AI that generates non-consensual intimate imagery or child sexual abuse material, and the date by which pre-existing generative systems must meet the Article 50(2) marking duty.
- 2 December 2027: high-risk obligations for Annex III use cases such as biometrics, education, employment and migration.
- 2 August 2028: high-risk obligations for AI embedded in regulated products listed in Annex I.
What the Omnibus changed
The Commission page describes the AI Omnibus as adopted on 19 November 2025, politically agreed on 7 May 2026 and in force from 27 July 2026. It moved the high-risk dates to the two later ones above, added the new prohibition, strengthened the AI Office, extended some SME relief to small mid-cap companies and broadened sandbox access.
If you planned around 2 August 2026 for high-risk compliance, the pressure has eased by about sixteen months for Annex III systems. That is not a reason to stop. Documentation, logging and risk management take a long time to build properly.
What still applies now
Several things are already live or live soon:
- Prohibited practices and AI literacy have applied since February 2025.
- General-purpose model providers have had obligations since August 2025. Models placed on the market before that date have until 2 August 2027.
- Transparency duties under Article 50 are tied to August 2026, with the December 2026 date for older generative systems.
A practical way to plan
Sort your AI use into three buckets. First, anything that could touch a prohibited practice: stop and review now. Second, anything user facing that chats, generates content or recognises emotion: transparency work. Third, anything in an Annex III area: start the inventory, documentation and oversight design, using the later dates as a runway rather than a finish line.
Keep a dated list of the sources you rely on, since Commission guidance, standards and codes of practice are still arriving.
Dates are also moving targets in one sense: guidance, templates and standards keep arriving even though the Act's own dates are set. Subscribe to Commission updates, and re-check the policy page before any customer facing statement.
Quick check
Did the Omnibus delay the whole AI Act?
No. It moved the high-risk dates for Annex III and Annex I systems. Prohibitions, AI literacy and general-purpose model duties kept their earlier dates.
When do the high-risk rules apply to a hiring tool?
Recruitment is an Annex III use, so the high-risk obligations apply from 2 December 2027 under the timeline above.
Sources
- European Commission, AI Act regulatory framework
- AI Act implementation timeline (an unofficial reader; confirm wording against the Official Journal)
Last reviewed: 19 October 2026.
