The AI Act timeline of obligations: what applies when
Adopted 2026-06-11 · ≈ 2 min read · 
edited by Dirk Baaijen
The AI Act takes effect in stages between 2025 and 2028. This timeline sets out which obligations apply on which date, and marks which dates are shifted by the Digital Omnibus agreement of 7 May 2026 — and which expressly are not.
Download the regime cheat sheet (PDF) ↓
The AI Act (Regulation (EU) 2024/1689) entered into force on 1 August 2024, but its obligations apply in stages. Since the political agreement on the Digital Omnibus of 7 May 2026, two timelines moreover run in parallel: the dates currently written in the regulation, and the dates politically agreed but not yet published in the Official Journal. Anyone navigating this field needs to know both.
What already applies
| Date | Obligation |
|---|---|
| 1 August 2024 | Entry into force of the regulation (Article 113). |
| 2 February 2025 | Prohibited AI practices (Article 5) and the AI literacy obligation (Article 4). |
| 2 August 2025 | Obligations for providers of general-purpose AI models (Chapter V), the governance structure (AI Office, AI Board) and the penalty regime. |
These dates are undisputed and are not rolled back by the Digital Omnibus agreement. The prohibitions and the literacy obligation have therefore applied for over a year, regardless of what happens to the high-risk dates.
What shifts: the high-risk obligations
The regulation itself names 2 August 2026 as the general date of application, including the obligations for high-risk systems under Annex III, and 2 August 2027 for high-risk systems embedded in regulated products (Annex I).
On 7 May 2026 the Council and the European Parliament reached a provisional political agreement on the Digital Omnibus, which shifts these dates:
| Subject | Current text | Political agreement 7 May 2026 |
|---|---|---|
| High-risk, Annex III (stand-alone systems) | 2 August 2026 | 2 December 2027 |
| High-risk, Annex I (in regulated products) | 2 August 2027 | 2 August 2028 |
| Transparency obligations, Article 50 | 2 August 2026 | 2 December 2026 |
The legal state of play
The agreement of 7 May 2026 is a provisional political agreement. Legal-linguistic revision, formal adoption by the Council and Parliament and publication in the Official Journal are still to follow. Until that publication, the current text of the regulation — with 2 August 2026 as the date of application — remains the law. Organisations that suspend their preparations in anticipation of the delay take a legal risk for as long as the amendment has not been published.
What this means in practice
Three sober conclusions. First: the obligations that already apply (prohibited practices, AI literacy, GPAI obligations) do not change and become nationally enforceable from 2 August 2026 through the designated supervisors. Second: the near-certain delay of the high-risk obligations gives organisations more time for the heaviest compliance building blocks — risk management system, data quality, technical documentation, conformity assessment — but changes nothing about their content. Third: anyone developing or procuring high-risk systems today is well advised to account for both timelines contractually, because the definitive dates are only fixed upon publication in the Official Journal.
This entry will be updated as soon as the amended regulation appears in the Official Journal.
Want these deadlines as a data feed for your own system? See API & data integration.
Sources
- https://eur-lex.europa.eu/eli/reg/2024/1689/oj
Regulation (EU) 2024/1689 (AI Act), authentic text in the Official Journal; Article 113 contains the application dates. - https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
European Commission policy page on the AI Act, with the current state of implementation. - https://www.hoganlovells.com/en/publications/eu-legislators-agree-to-delay-for-highrisk-ai-rules
Analysis of the political agreement of 7 May 2026 delaying the high-risk obligations (Digital Omnibus). - https://www.aiactblog.nl/en/digital-omnibus
Continuously updated overview of the Digital Omnibus amendments and the status of formal adoption. - https://artificialintelligenceact.eu/implementation-timeline/
Unofficial but widely used implementation tracker by the Future of Life Institute.
Read next
The state of AI regulation — 2026 overview
One overview of where AI regulation stands in 2026: the phased AI Act, the shifts from the Digital Omnibus, the five regimes, and the international line from the Council of Europe to California and Korea — with links to the detail per topic.
Right to explanation of an AI decision: what Article 86 of the AI Act gives you
If you are affected by a decision based (in part) on a high-risk AI system, Article 86 of the AI Act gives you the right to a clear explanation of the AI system's role and the main elements of the decision — from the deployer, on top of your GDPR rights.
DPIA for HR AI: when is it mandatory and how do you combine it with the FRIA?
A DPIA (Art. 35 GDPR) is mandatory for large-scale, systematic monitoring and for high-risk AI in HR. This article explains what it must contain and how to combine the DPIA with the FRIA (fundamental rights assessment, Art. 27 AI Act) into one process. With a practical step plan.