Skip to main content

Implementation timeline

EU AI Act timeline: when each layer starts to apply

Follow the staged application dates in Regulation (EU) 2024/1689. The timeline separates dates in enacted law from implementation instruments and proposals that may still change.

Binding dates and implementation steps

Dates in chronological order

Past milestones remain visible for context.

  1. Passed

    AI Act enters into force

    Regulation (EU) 2024/1689 published in the Official Journal on 12 July 2024.

  2. Passed

    Prohibited AI practices apply

    Art. 5 bans (e.g. social scoring, manipulative techniques, real-time remote biometric identification in public spaces) become enforceable.

  3. Passed

    GPAI obligations · AI Office governance

    Chapters III (high-risk) and V (GPAI models) apply for general-purpose AI; AI Office, Board and scientific panel operational; penalties for prohibited AI.

  4. Passed

    High-risk AI (Annex III) applies

    Most Annex III high-risk use cases and transparency obligations for certain AI (chatbots, emotion recognition) become applicable.

  5. Upcoming

    Full application — Annex I product safety

    High-risk AI embedded in products under Union harmonisation legislation (Annex I) and all remaining obligations fully applicable.

Official legal sources

AML Map is an orientation layer. Use the consolidated legal text and the responsible authority's publications for decisions and filing deadlines.

Turn dates into next actions

Continue from the timeline

Find your role

Start with provider, deployer, importer, distributor, GPAI, or an Annex III use case.

Open

Track implementation rules

Review guidelines, templates, standards, common specifications, and codes by status.

Open

Watch consultations

Review open and upcoming Commission and AI Office consultation dates.

Open

Digital Omnibus — proposed AI Act timeline changes

COM(2025) 836 proposes a stop-the-clock mechanism: high-risk obligations would apply only once harmonised standards and official guidelines are available, with backstop dates if those instruments are delayed.

Under negotiation between the European Parliament and the Council (Feb 2026).

  • Annex III high-risk use cases: proposed backstop 2 December 2027 (statutory 2 August 2026)
  • Annex I product-safety embedded AI: proposed backstop 2 August 2028 (statutory 2 August 2027)

COM(2025) 836 — full proposal · AML & classification guide