Digital Omnibus on AI — Regulation (EU) 2026/1744 and the AI Act amendments
Regulation (EU) 2026/1744 (Digital Omnibus on AI) was published on 24 July 2026 and entered into force on 27 July 2026. It defers high-risk AI obligations, adds two new prohibited practices and softens the AI literacy duty. Article 50 transparency obligations were NOT deferred.
Source documentWhat the Digital Omnibus on AI is
Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amends Regulation (EU) 2024/1689 (the AI Act), Regulation (EU) 2018/1139 and Regulation (EU) 2023/1230.
| Published in the Official Journal | 24 July 2026 (OJ L, 2026/1744) |
| Entry into force | 27 July 2026 |
| CELEX | 32026R1744 |
| ELI | http://data.europa.eu/eli/reg/2026/1744/oj |
This is the only amendment to the AI Act adopted to date.
A note on naming. The label "Omnibus VII" circulates in media coverage and internal materials. It is not the name of any legal act — it is an informal label for the European Commission's seventh simplification package. The official name is the Digital Omnibus on AI, and the legal source is Regulation (EU) 2026/1744. Cite the regulation number, not "Omnibus VII".
Deadlines after the amendment
| Obligation | Applies from | Amended? |
|---|---|---|
| Art. 5 prohibited practices, points (a)–(h) | 2 February 2025 | no |
| Art. 4 AI literacy | 2 February 2025 | date no, wording yes |
| GPAI obligations (Chapter V) | 2 August 2025 | no |
| Art. 50 transparency obligations | 2 August 2026 | no |
| Art. 50(2) marking — systems placed on the market before 2 Aug 2026 | 2 December 2026 | new Art. 111(4) |
| Art. 5(1) points (ba) and (bb) — new prohibitions | 2 December 2026 | new |
| Annex III standalone high-risk | 2 December 2027 | yes, from 2 Aug 2026 |
| Annex I embedded high-risk | 2 August 2028 | yes, from 2 Aug 2027 |
| GPAI models placed on the market before 2 Aug 2025 | 2 August 2027 | no |
| Regulatory sandboxes operational | 2 August 2027 | yes |
| High-risk systems in use by public authorities | 2 August 2030 | no |
Article 50 was not deferred — a common misreading
Transparency obligations apply from 2 August 2026, unchanged. This covers:
- Art. 50(1) — providers must ensure users know they are interacting with an AI system
- Art. 50(3) — deployers inform people exposed to emotion recognition or biometric categorisation
- Art. 50(4) — deployers label deep fake content and AI-generated text published to inform the public
The only transitional relief comes from the new Art. 111(4) and it is narrow. It applies only where all three conditions are met:
- you are a provider, not a deployer,
- the obligation at issue is Art. 50(2) only (machine-readable marking and detectability of outputs),
- the system was placed on the market before 2 August 2026.
In that case the deadline is 2 December 2026. A deployer using a deep fake system from 2 August 2026 gets no transitional period at all, regardless of when the system was placed on the market.
The high-risk deferral is unconditional
The dates 2 December 2027 and 2 August 2028 are fixed calendar dates. The Commission's proposal COM(2025) 836 contained a mechanism under which application would depend on the Commission confirming the availability of harmonised standards — that mechanism was not adopted. Do not plan on the assumption that the deadlines may shift again.
New prohibited practices — Art. 5(1)(ba) and (bb)
The regulation inserts two new points after the existing point (b), applying from 2 December 2026:
- (ba) — AI systems generating or manipulating realistic images, video, audio or similar material depicting an identifiable natural person's intimate parts, or that person engaged in sexually explicit activities, without their freely given, specific, informed, unambiguous and explicit consent.
- (bb) — the same, for material or a performance within the meaning of Article 2, points (c) and (e), of Directive 2011/93/EU (child sexual abuse material), except where a "without right" defence applies under national law.
Two new paragraphs that substantially narrow the scope were added alongside them (Art. 5(1a) and 5(1b)), with the same application date. Covered in detail in a separate article: The Art. 5(1)(ba) and (bb) prohibition.
Other amendments
- Art. 4 softened — the wording moved from "ensure a sufficient level" to "take measures to support the development" of AI literacy, with an explicit statement that the obligation does not require guaranteeing any specific level of literacy for any individual. The new wording applies from 27 July 2026. The obligation itself has existed since 2 February 2025.
- "Safety component" narrowed — excludes systems for user assistance, optimisation and quality control, unless failure endangers health or safety.
- Art. 2(13) — the Commission may limit duplicative requirements by delegated act where sectoral legislation provides equivalent safeguards; the deadline for adoption is 2 August 2027.
- Registration for self-assessed non-high-risk systems (Art. 6(3)) is retained but simplified.
- Art. 50(7) — the Commission's power to approve codes of practice by implementing act was removed. The substantive obligations in Art. 50(1)–(6) are untouched.
Limits of this overview
- A consolidated official text of the AI Act with the amendments incorporated does not yet exist. This overview is a reconstruction from the two source Official Journal texts (2024/1689 and 2026/1744). Provision numbering may be presented differently in a future consolidated version.
- The date from which the new prohibitions in (ba) and (bb) become enforceable is not expressly stated. Chapter XII (penalties) has applied since 2 August 2025, while the prohibitions themselves apply only from 2 December 2026. Do not draw conclusions about penalties without legal advice.
- Quotations are from the English version of the Official Journal.
artificialintelligenceact.euand the European Commission's AI Act Service Desk had not been updated as at the date of this article and still show pre-omnibus deadlines. The Service Desk carries its own notice to that effect on Art. 113.
Sources
Need compliance documentation?
Generate AI Inventory, Risk Assessment and other documents automatically — tailored to your system.