reportEU AI Act

The EU AI Act so far: every milestone from adoption to mid-2026

A dated, primary-sourced timeline infographic of Regulation (EU) 2024/1689: entry into force, the phased application dates and the Digital Omnibus deferrals.

Edited and verified by Cognesio LLP · updated

Researched with AI assistance · sources verified by Cognesio LLP · How this was made ↓

  1. European Parliament adopts the Act (523 to 46)
  2. Published in the Official Journal (OJ L, 2024/1689)
  3. Entry into force (Art. 113)
  4. Prohibited practices (Art. 5) and AI-literacy duty (Art. 4) apply
  5. GPAI model obligations, governance and penalties apply
  6. Digital Omnibus on AI (Reg. 2026/1744) enters into force; high-risk dates deferred
  7. Article 50 transparency obligations apply; Commission enforcement over GPAI providers begins
  8. New prohibitions in Art. 5(1)(ba) and (bb) apply; Art. 50(2) marking for existing generative systemsupcoming
  9. Annex III high-risk obligations apply (deferred from 2 August 2026)upcoming
  10. Annex I high-risk obligations, AI in regulated products (deferred from 2 August 2027)upcoming
  11. High-risk systems for public authorities already on the market must complyupcoming
EU AI Act timeline, as amended by the Digital Omnibus on AI (Regulation (EU) 2026/1744). Dates from the tables below; verified against EUR-Lex on 3 September 2026.

Updated 3 September 2026: the Digital Omnibus on AI (Regulation (EU) 2026/1744) was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. The deferred high-risk dates are now law. The “adopted, not yet in force” caveat below described the position as of 15 July 2026 and is kept as written; the new rows are in the “Summer 2026” section. Monthly detail is in the September 2026 update.

The EU AI Act (Regulation (EU) 2024/1689) is the first comprehensive statutory regime for artificial intelligence. It entered into force on 1 August 2024 and applies in phases. As of 15 July 2026, a simplification package known as the Digital Omnibus has been adopted by the EU co-legislators but has not yet been published in the Official Journal — so it is not yet in force, and the Act’s original application dates still govern.

That distinction matters for anyone planning compliance: what has been agreed and what is legally operative are, for now, two different things.

The core timeline

DateMilestoneSource
13 March 2024European Parliament adopts the Act (523–46)Article 113 regime
21 May 2024Council of the EU gives final approval—
12 July 2024Published in the Official Journal (OJ L, 2024/1689)EUR-Lex
1 August 2024Entry into force (Art. 113)Art. 113
2 February 2025Prohibited practices (Art. 5) and AI-literacy duty (Art. 4) applyArt. 113(a)
2 August 2025GPAI model obligations, governance, and penalties applyArt. 113(b)
2 August 2026General application date — most high-risk (Annex III) obligations. Superseded: deferred to 2 December 2027 by Reg. 2026/1744 (in force 27 July 2026)Art. 113 chapeau; amended Art. 113(c)(i)
2 August 2027Extended transition for high-risk AI in regulated products. Superseded: deferred to 2 August 2028 by Reg. 2026/1744Art. 113(c); amended Art. 113(c)(ii)

Guidelines and codes

Two interpretive instruments sit alongside the statute. On 4 February 2025 the Commission published non-binding guidelines on the Article 5 prohibited practices. On 10 July 2025 the AI Office published the General-Purpose AI Code of Practice — a voluntary tool covering transparency, copyright, and safety and security — to support compliance with the Article 53 and 55 obligations for GPAI models.

Neither is binding law. Authoritative interpretation of the Act rests with the Court of Justice.

The Digital Omnibus: adopted, not yet in force

In late 2025 the Commission moved to defer parts of the timetable. The sequence:

  • 19 November 2025: the Commission publishes the Digital Omnibus on AI, a legislative proposal to simplify and postpone certain obligations.
  • 7 May 2026: Parliament and Council reach a provisional agreement in trilogue.
  • 16 June 2026: the European Parliament adopts the agreed text.
  • 29 June 2026: the Council gives its final approval, completing co-legislator adoption.

As adopted, the Omnibus would defer the high-risk application dates to 2 December 2027 for stand-alone Annex III systems and 2 August 2028 for high-risk AI embedded in regulated products, and would add a new prohibited practice targeting AI-generated child sexual abuse material and non-consensual intimate imagery, with compliance envisaged by 2 December 2026.

The caveat is decisive. As of 15 July 2026 the amending regulation had not yet been signed and published in the Official Journal. Until that publication, Article 113’s original dates, including 2 August 2026 for general application, remain the operative legal text. The deferrals are agreed, not yet effective.

What to watch

The next hard event is the Official Journal publication of the Digital Omnibus, which will fix the revised dates and bring the new prohibition into the timetable. From there, the monthly EU AI Act updates on this site track enforcement and guidance as they land.

Summer 2026: the Omnibus in force and Article 50 live

Added 3 September 2026. The Official Journal publication anticipated above happened on 24 July 2026. The rows below are verified against EUR-Lex and the Commission’s pages on 3 September 2026; the amended Article 113 text is quoted from the Omnibus as published.

DateMilestoneSource
8 July 2026Commission opinion finds the Code of Practice on Transparency of AI-generated Content adequate for Articles 50(2), (4) and (5); AI Board adequacy assessment follows on 9 JulyCommission opinion page
8 July 2026Regulation (EU) 2026/1744 (Digital Omnibus on AI) signed at StrasbourgEUR-Lex, Reg. 2026/1744
20 July 2026Commission approves final Guidelines on the Article 50 transparency obligations, C(2026) 5054 finalCommission library page
24 July 2026Regulation (EU) 2026/1744 published in the Official Journal (OJ L, 24.7.2026)EUR-Lex
27 July 2026Regulation (EU) 2026/1744 enters into force (Art. 4: third day after publication); amended Art. 113(d): Articles 102 to 110 applyEUR-Lex; Commission news, 27 July 2026
2 August 2026Article 50 transparency obligations apply; Commission enforcement powers over GPAI providers apply; AI Office begins enforcingArt. 113; Commission news, 31 July 2026
2 December 2026New prohibitions in Art. 5(1)(ba) and (bb) apply; providers of generative systems on the market before 2 August 2026 must meet Art. 50(2) markingAmended Art. 113(a); new Art. 111(4)
2 August 2027Commission delegated acts under new Art. 2(13) dueReg. 2026/1744, point (3)
2 September 2027Commission post-market monitoring guidance and template dueAmended Art. 72(3)
2 December 2027Chapter III, Sections 1 to 3 apply to Annex III high-risk systems (Art. 6(2))Amended Art. 113(c)(i)
2 August 2028Chapter III, Sections 1 to 3 apply to Annex I high-risk systems (Art. 6(1))Amended Art. 113(c)(ii)
2 August 2030High-risk systems intended for public authorities and already on the market must complyAmended Art. 111(2)

Article 4 was also replaced: providers and deployers now “take measures to support the development of AI literacy” and the text states the obligation “does not require providers or deployers to guarantee any specific level of AI literacy of any individual”. Its application date, 2 February 2025, did not move.

Sources

Reuse this research

Quote it, cite it, forward it — CC BY 4.0 for the data and figures; the linked official documents are the record. Suggested citation and a link to this exact report: