Regulation 2024/1689 · Application dates

EU AI Act Timeline

When each part of the Act applies, after the Digital Omnibus. The dates come from the law itself, so this page changes only if the Act is amended again. For guidelines, standards and enforcement news, follow the Commission's official timeline and the EU AI Act Newsletter.

Dates as set byRegulation 2024/1689 as amended by Regulation 2026/1744 (OJ, 24 Jul 2026)
42daysuntil next milestone
5/ 10milestones in force
79controlsmapped in blueprint →
€35M / 7%max for prohibited practices

Enforcement timeline

In forceUpcoming
  1. In force

    AI Act enters into force

    Regulation 2024/1689 published in Official Journal, 20-day entry into force.

  2. In force

    Prohibited practices + AI literacy

    Art. 5 prohibitions (social scoring, subliminal manipulation, real-time biometric ID) and Art. 4 AI literacy obligation take effect.

  3. In force

    GPAI model obligations + governance

    Art. 51-56 GPAI provider obligations (technical docs, copyright, downstream info). AI Office and governance structure operational.

  4. In force

    Digital Omnibus on AI entered into force

    Regulation (EU) 2026/1744, published in the Official Journal 24 July 2026, in force the third day after. First formal amendment to the AI Act since adoption.

  5. In force

    Transparency obligations (Art. 50)

    Chatbot disclosure, deepfake/AI-content labelling, emotion-recognition and biometric notification now apply (final version of the Commission guidelines published 20 July 2026; the voluntary Code of Practice on marking and labelling AI-generated content, final since 10 June 2026, gives signatories a way to show compliance with Art. 50(2), (3) and (5)). The Commission can now fine GPAI providers (Art. 101), and the AI Office supervises AI systems built on a provider's own GPAI model (Art. 75(1) as amended). Providers of generative AI already on the market have until 2 Dec 2026 for the Art. 50(2) marking duty.

  6. You are here · 21 Oct 2026
  7. Upcoming42d

    Watermarking + nudifier ban

    Providers of generative AI systems placed on the market before 2 Aug 2026 must meet the Art. 50(2) machine-readable marking duty. New prohibitions on AI that generates non-consensual intimate imagery or CSAM apply.

    Both added by the Digital Omnibus (Regulation (EU) 2026/1744): Art. 111(4) for the marking transition, Art. 5(1)(ba)-(bb) applying from this date under Art. 113.

  8. Upcoming285d

    Regulatory sandboxes

    Deadline for each Member State to have at least one national AI regulatory sandbox operational (Art. 57(1)).

    Moved from 2 Aug 2026 to 2 Aug 2027 by the Digital Omnibus (Regulation (EU) 2026/1744).

  9. Upcoming407d

    High-risk AI systems — Annex III

    Full Art. 8-15 requirements for standalone high-risk AI systems: employment, credit scoring, law enforcement, education, critical infrastructure, biometrics, migration, justice.

    Postponed from 2 Aug 2026 to 2 Dec 2027 by the Digital Omnibus (Regulation (EU) 2026/1744, in force 27 July 2026).

  10. Upcoming651d

    High-risk AI in regulated products — Annex I

    AI systems that are products or safety components under the EU product laws in Annex I Section A: medical devices, in vitro diagnostics, toys, lifts, radio and pressure equipment. Vehicles, aviation, marine and rail (Section B) follow their own sectoral rules (Art. 2(2)).

    Postponed from 2 Aug 2027 to 2 Aug 2028 by the Digital Omnibus (Regulation (EU) 2026/1744). AI in machinery moved to a sectoral track, with AI-specific requirements to follow via delegated acts by 2 Aug 2028.

  11. Upcoming1532d

    Large-scale IT systems (Annex X)

    Legacy AI systems in large-scale EU IT systems (SIS, VIS, Eurodac, etc.) must be brought into compliance.

Go deeper

EU AI Act Blueprint →79 controls, 73 of them tied to a specific article of the Act. Open source on GitHub.Official timeline →The Commission's AI Act Service Desk: guidelines, standards and implementing acts as they are adopted.