EU AI Act deadlines: the post-Omnibus timeline

Every statutory application date of Regulation (EU) 2024/1689 as amended by the Digital Omnibus — verified against official sources, with the legal status of each date stated explicitly.

The Digital Omnibus on AI was signed on 8 July 2026 and is awaiting publication in the Official Journal; it enters into force three days after publication. Dates marked below take effect as amended once published.

Last verified: 2026-07-19

  1. In force

    Prohibited AI practices and AI-literacy duties

    Article 5 prohibitions (social scoring, exploitative manipulation, untargeted facial-image scraping, emotion recognition at work and in education, and related practices) and the Article 4 AI-literacy obligation apply. These were unchanged by the Digital Omnibus and are enforceable now.

    Sources: Regulation (EU) 2024/1689, Art. 113 (EUR-Lex)

  2. In force

    GPAI-model obligations, governance and penalties

    Chapter V obligations for general-purpose AI models, the governance framework including the AI Office (Chapter VII), and the penalties regime (Chapter XII) apply. Member States were also due to designate national market-surveillance and notifying authorities by this date.

    Sources: Regulation (EU) 2024/1689, Art. 113 (EUR-Lex)

  3. Applies from 2 August 2026

    Article 50 transparency obligations

    Chatbot disclosure, deepfake and synthetic-content marking, and biometric-system notices apply. Under the Digital Omnibus this date now carries only these transparency obligations — the Annex III high-risk obligations originally due today move to 2 December 2027. Watermarking applies to newly placed systems; systems already on the market have until 2 December 2026.

    Sources: Regulation (EU) 2024/1689, Art. 113 (EUR-Lex) · What actually applies on 2 August 2026 (European Express, 17 Jul 2026)

  4. Applies from 2 December 2026Signed — pending OJ publication

    Watermarking grace period ends; NCII prohibition

    The grace period for watermarking compliance by AI systems already on the market before 2 August 2026 ends, and the new prohibition on AI-generated non-consensual intimate imagery takes effect. Both were introduced by the Digital Omnibus.

    Sources: Council final green light, 29 Jun 2026 (Consilium)

  5. Applies from 2 August 2027Signed — pending OJ publication

    National AI regulatory sandboxes

    Deadline for national competent authorities to establish AI regulatory sandboxes, postponed by one year under the Digital Omnibus.

    Sources: Council final green light, 29 Jun 2026 (Consilium)

  6. Applies from 2 December 2027Signed — pending OJ publication

    Stand-alone high-risk AI obligations (Annex III)

    Obligations for stand-alone high-risk AI systems — recruitment and worker management, credit scoring, education, access to essential services, law enforcement, border control and critical infrastructure — apply. This is the headline change of the Digital Omnibus: a 16-month postponement from the original 2 August 2026 date.

    Previously: 2 August 2026. Postponed 16 months by the Digital Omnibus (signed 8 Jul 2026, pending OJ publication).

    Sources: Council final green light, 29 Jun 2026 (Consilium) · Omnibus agreement summary (Gibson Dunn)

  7. Applies from 2 August 2028Signed — pending OJ publication

    High-risk AI embedded in regulated products (Annex I)

    Obligations for high-risk AI embedded in products already covered by EU product-safety law — medical devices, machinery, toys, vehicles — apply, postponed one year from the original 2 August 2027 date by the Digital Omnibus.

    Previously: 2 August 2027. Postponed 12 months by the Digital Omnibus (signed 8 Jul 2026, pending OJ publication).

    Sources: Council final green light, 29 Jun 2026 (Consilium)

Reuse this table: the milestone data on this page is licensed CC BY 4.0 — cite “PrivacyForge EU AI Act deadline tracker” with a link.

Which risk tier is your AI system in?

Answer three questions about what the system does. You get the likely EU AI Act risk tier, the obligations that follow, and the deadline that applies — no signup needed.

Does the system do any of the following?

These practices are prohibited outright under Article 5 — in force since 2 February 2025.

None of these apply

Frequently asked questions

When does the EU AI Act fully apply?
In stages. Prohibitions and AI-literacy duties have applied since 2 February 2025, and GPAI-model, governance and penalty provisions since 2 August 2025. Article 50 transparency obligations apply from 2 August 2026. Under the Digital Omnibus, stand-alone high-risk (Annex III) obligations apply from 2 December 2027 and high-risk AI embedded in regulated products (Annex I) from 2 August 2028.
What did the Digital Omnibus change?
It postponed the high-risk deadlines: Annex III obligations moved from 2 August 2026 to 2 December 2027, and Annex I obligations from 2 August 2027 to 2 August 2028. It also added a watermarking grace period ending 2 December 2026 for systems already on the market, and a new prohibition on AI-generated non-consensual intimate imagery from the same date. It did not change the prohibitions, AI-literacy, GPAI or penalty provisions already in force.
Is the postponement law yet?
Effectively yes, with one formality outstanding. The European Parliament adopted the Digital Omnibus on 16 June 2026, the Council gave its final green light on 29 June 2026, and the final act was signed on 8 July 2026. It enters into force three days after publication in the Official Journal, which is pending. This page states the pending status on every affected date and is updated when publication is confirmed.
What actually applies on 2 August 2026?
Only the Article 50 transparency obligations: telling people when they interact with an AI system such as a chatbot, disclosing and machine-readably marking synthetic content and deepfakes, and informing people when emotion recognition or biometric categorisation is used. The Annex III high-risk obligations originally due on this date move to 2 December 2027 under the Digital Omnibus.
Does the AI Act apply to eCommerce personalisation and chatbots?
Usually as limited or minimal risk. Customer-facing chatbots and AI-generated product content trigger Article 50 transparency duties from 2 August 2026. Recommendation engines and search ranking are typically minimal risk. Emotion inference in the workplace, social scoring, and certain biometric uses are prohibited outright, and uses such as AI-driven credit scoring or recruitment screening are high-risk under Annex III.
Is this page legal advice?
No. It is a maintained summary of the statutory application dates with links to the official sources for every entry. For decisions with legal consequences, confirm the details against the cited sources and a qualified advisor.

New to the topic? Start with our guide to AI governance and the EU AI Act compliance guide.