2026-06-29 - Digital Omnibus on AI: the Council's Final Green Light
On 29 June 2026, the Council of the European Union gave its final green light to the Digital Omnibus on AI, formally adopting the text agreed with the European Parliament and approved by the latter on 16 June 2026. With this step, the ordinary legislative procedure on file 2025/0359(COD) comes to a close.
This article does not repeat the article-by-article analysis, which I already provided in the piece on the consolidated compromise text and recalled in the one on the Parliament’s approval; I refer to those pieces for the regulatory detail. Here, I examine the legal meaning of the Council’s adoption. The Council does not change the already agreed content today: rather, it makes the political-institutional move toward a new AI Act calendar irreversible. The legal dividing line remains publication in the Official Journal, but for operators and authorities the period of regulatory uncertainty is now over.
The Council’s final green light
The Council’s adoption, within the dynamics of the first-reading agreement (Article 294 TFEU), represents the Council’s approval of the same text already voted on by the Parliament. Because the two institutions have approved the identical wording, the act is adopted without further readings.
The Council frames the measure as part of the simplification package, with the stated objective of reducing recurring administrative costs for businesses and ensuring a more harmonized and proportionate application of the AI rules across the Union. In my view, what matters to record is not the “simplifying” reach as such — already examined in the previous pieces — but the fact that adoption arrives just weeks before the AI Act’s first application deadline.
From agreed text to positive law: signature, Official Journal, entry into force
The Council’s adoption exhausts the co-legislators’ ordinary decision-making stage; publication in the Official Journal, however, remains the step that turns the text into positive law. Two steps remain: the signature by the presidents of the two institutions and publication in the Official Journal of the European Union.
Until publication in the Official Journal and the entry into force of the amending regulation, the AI Act’s text in force and its original application calendar continue to apply. The regulation will enter into force on the third day following its publication in the Official Journal.
On the technical plane, it is worth specifying the act’s scope. The amending regulation applies not only to Regulation (EU) 2024/1689 (the AI Act) and Regulation (EU) 2018/1139 (the EASA Regulation), but also to Regulation (EU) 2023/1230 (the Machinery Regulation). The reference to the Machinery Regulation reflects a specific choice of the compromise: for products falling within its scope, the direct application of the AI Act is excluded, and the Commission may introduce, within the framework of the Machinery Regulation, health and safety requirements relating to AI systems classified as high-risk under the AI Act.
What remains unchanged relative to the text already approved by Parliament
Relative to the agreed text, the Council’s green light changes nothing: the architecture confirmed by Coreper on 13 May 2026, approved by the IMCO and LIBE committees on 2 June 2026, and voted by the Parliament on 16 June is the same one the Council adopts today. For the detail — the new calendar, the Article 5 prohibitions, AI literacy (Article 4), the processing of special categories of data for bias detection and correction, the AI Office’s powers, the sandboxes, the transparency obligations, and the coordination with sectoral legislation — I refer to the earlier articles in the series.
A misunderstanding must, however, be avoided: “unchanged” holds relative to the articles I have already published, not relative to the law in force. Relative to the latter, the regulation changes significantly. It moves the application of the obligations for high-risk AI systems to 2 December 2027 for those in Annex III and to 2 August 2028 for those embedded as safety components in Annex I products; and it introduces, from 2 December 2026, the new prohibitions.
A brief substantive recall is appropriate, because the Council expressly emphasizes the content as well, not only the timing: the ban on non-consensual sexual or intimate content and on child sexual abuse material (CSAM) generated or manipulated by AI; the postponement of the deadline for ensuring the operation of at least one regulatory sandbox at national level; the reduction of the transitional period for certain transparency solutions on artificially generated content; the clarifications on the AI Office’s competences vis-à-vis national authorities; and the mechanisms to resolve overlaps with sectoral legislation. For the detailed analysis of each profile, I refer again to the pieces on the consolidated text and on the Parliament’s approval.
.tbl-wrap{overflow-x:auto;margin:1.5rem 0}
.tbl-wrap table.compare{width:100%;border-collapse:collapse;font-size:.92rem;line-height:1.45}
.tbl-wrap table.compare th{background:#26215c;color:#fff;padding:.6rem .7rem;text-align:left;vertical-align:top}
.tbl-wrap table.compare td{padding:.55rem .7rem;border:1px solid #d9d9e3;vertical-align:top}
.tbl-wrap table.compare tr:nth-child(even) td{background:#f4f4fa}
.tbl-wrap table.compare td:first-child{font-weight:600;white-space:nowrap}
.tbl-note{font-style:italic;font-size:.85rem;margin-top:.4rem}
@media(prefers-color-scheme:dark){
.tbl-wrap table.compare td{border-color:#3a3a4a}
.tbl-wrap table.compare tr:nth-child(odd) td{background:#16161f}
.tbl-wrap table.compare tr:nth-child(even) td{background:#1e1e2a}
.tbl-wrap table.compare td:first-child{color:#e8e8f2}
}
DateScope
2 December 2026New Article 5 prohibitions concerning non-consensual sexual or intimate content and child sexual abuse material (CSAM); compliance with the machine-readable marking obligations under Article 50(2) for systems placed on the market before 2 August 2026
2 August 2027Operation of at least one regulatory sandbox at national level by the competent authorities
2 December 2027Stand-alone high-risk AI systems — Annex III
2 August 2028High-risk AI systems embedded as safety components in products covered by sectoral legislation — Annex I
The dates set out above are established by the regulation adopted by the Council. The regulation will enter into force on the third day following its publication in the Official Journal; from that moment, the new calendar will replace the existing one. The individual provisions will apply on the respective dates indicated in the table.
Why the step matters on the eve of 2 August 2026
The compression of the timeline has a precise reason. 2 August 2026 is the date on which, under the text in force, the obligations for Annex III high-risk systems would begin to apply. The stated objective is to complete the procedure — adoption, signature, and publication — before that date, to replace the existing calendar with the new deadlines before the first term falls due.
In my view, it is precisely this time window that explains the speed of the entire procedure, from the 19 November 2025 proposal to the late-June adoption. A margin of operational risk remains, however: the substitution effect of the new calendar takes place only upon entry into force, and therefore after publication in the Official Journal. Until that happens, anyone planning exclusively against the new dates is doing so de lege ferenda.
The immediate operational implications: distinguishing law in force from the calendar to come
On the operational level, two planes must be kept distinct. The first is the law in force: until the amending regulation enters into force, the AI Act applies in its current wording and according to its original calendar; the Article 5 prohibitions already in force, the AI literacy obligation under Article 4, and the obligations on GPAI models — applicable since 2 August 2025 — remain fully applicable.
The second is the calendar to come: providers, deployers, DPOs and public administrations may reasonably begin planning against the new dates, with the awareness that they will become binding only when the regulation enters into force, following its publication in the Official Journal. In my view, the professionally sound course of action is to begin now with the inventory and classification of systems — activities that do not depend on the availability of harmonised standards — and to treat the new deadlines as a planning horizon, not as dates already in force. To navigate the AI Act’s regulatory ecosystem, I refer to the permanent reference pages of this blog listed below.
Related resources on this blog
This article closes the series devoted to the Digital Omnibus, which has followed the file from the consultation stage through to the Council’s adoption:
The EU Digital Omnibus: Europe’s Bold Move to Simplify Digital Regulation
Digital Omnibus on AI: European Commission Proposes Simplifications to the AI Act
Digital Omnibus: Cookies, GDPR and AI Training — New European Privacy Rules
Digital Omnibus on AI: the European Parliament Rewrites the Commission’s Rules
Digital Omnibus on AI: Council and Parliament Align Mandates as Trilogue Negotiations Begin
Digital Omnibus on AI: EP Adopts Position (569 Votes)
Digital Omnibus on AI: the Provisional Agreement of 7 May 2026
Digital Omnibus on AI: the consolidated compromise text heading to the plenary vote
Digital Omnibus on AI: the European Parliament Approves the Agreed Text
For a systematic orientation within the AI Act regulatory ecosystem, the following permanent reference pages are also available:
- AI Rules Hub — map of the AI Act and related legislation
- AI Operators Hub — obligations for providers, deployers and other operators
- AI–GDPR Hub — intersections between the AI Act and data protection
Sources
Council of the European Union, press release, Artificial intelligence: Council gives final green light to simplify and streamline rules, 29 June 2026 —
European Commission, proposal for a regulation COM(2025) 836 (Digital Omnibus on AI), 19 November 2025 —
https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52025PC0836
Procedure file 2025/0359(COD) — Legislative Observatory —
https://oeil.europarl.europa.eu/oeil/en/procedure-file?reference=2025/0359(COD)
European Parliament, Legislative Train Schedule — Digital Omnibus on AI —
https://www.europarl.europa.eu/legislative-train/package-digital-package/file-digital-omnibus-on-ai
Regulation (EU) 2024/1689 (AI Act) — EUR-Lex ELI —
https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng
Regulation (EU) 2018/1139 (EASA Regulation) — EUR-Lex ELI —
https://eur-lex.europa.eu/eli/reg/2018/1139/oj/eng
Regulation (EU) 2023/1230 (Machinery Regulation) — EUR-Lex ELI —
https://eur-lex.europa.eu/eli/reg/2023/1230/oj/eng
New to Gemini? This capsule is best read with a Gemini client, such as Lagrange.