On 27 July 2026, Regulation (EU) 2026/1744, the amending act the Commission calls the AI Omnibus, entered into force across the Union. [1] [2] It is the AI Act portion of the Digital Omnibus package the Commission proposed on 19 November 2025. [1] The Commission describes the regulation as targeted amendments to the EU AI Act, adopted in June 2026. [3]
The amended calendar moves the heavy high-risk work into December 2027 and August 2028, keeps every obligation already in force intact, and lands two new prohibitions and one marking deadline on 2 December 2026, 26 days from this article's date of 6 November 2026. Companies planning European AI compliance against the pre-amendment calendar are working from a superseded schedule. A program that targeted August 2026 now has until December 2027, and the nearest new deadline falls inside 2026.
The high-risk dates that moved
Two sets of high-risk obligations now apply later than the 2024 text said. The Commission's entry-into-force announcement puts the rules for high-risk AI systems in the Annex III categories at 2 December 2027, and the rules for high-risk AI embedded in regulated physical products such as machinery and lifts at 2 August 2028. [1] euaiact.com, a vendor-maintained AI Act resource, lists the eight Annex III use-case areas as biometrics, critical infrastructure, education, employment, essential services, law enforcement, migration and the administration of justice. [4] The authoritative enumeration is Annex III of the Official Journal text of the AI Act; the vendor's labels are compressed paraphrases, and its own FAQ introduces the list with "including". [4] The AI Act Service Desk implementation timeline lists both dates under the amendment marker and sets full roll-out of the main milestones by 2 August 2028. [5]
The 2024 text had set 2 August 2026 for the Annex III systems and 2 August 2027 for the embedded products. The same resource records the move as a 16-month deferral for Annex III and a 12-month deferral for Annex I, and attributes it to harmonised standards and notified-body infrastructure that were not ready. [6]
| Obligation | Original date | Now applies |
|---|---|---|
| High-risk AI systems, Annex III use cases | 2 August 2026 | 2 December 2027 |
| High-risk AI embedded in regulated products (Annex I) | 2 August 2027 | 2 August 2028 |
New dates are stated by the Commission. [1] [5] Original dates and deferral lengths come from the vendor resource. [6]
The obligations already in force
Nothing already in force was rolled back. [6] The Service Desk timeline shows prohibitions and the AI literacy duty applying since 2 February 2025, and general-purpose AI model obligations applying since 2 August 2025 alongside national penalty laws and EU-level governance bodies. [5] From 2 August 2026, the Article 50 transparency rules apply, and enforcement starts at national and EU level over general-purpose AI models, prohibitions, transparency rules and AI literacy. [5]
Reach is not limited to EU-incorporated companies. The euaiact.com overview states that the Act applies to providers placing systems on the EU market regardless of where they are established, and to providers and deployers outside the Union where the system's output is used in the EU. [4] A SaaS vendor serving customers in the EU from outside the bloc is inside that description.
The deadlines landing on 2 December 2026
The nearest dates in the amended calendar are additions made by this amendment. The timeline states that prohibitions covering AI systems that generate non-consensual sexual deepfakes and child sexual abuse material start to apply on 2 December 2026. [5] The Commission's announcement describes the same change as a ban on nudification apps, prohibiting systems that generate non-consensual sexually explicit and intimate content or child sexual abuse material. [1] euaiact.com places the two bans in new Article 5(1)(ba) and Article 5(1)(bb) and counts them in the penalty tier with a maximum of €35 million or 7 percent of worldwide annual turnover. [6] [4]
The same date sets a transitional deadline for certain providers on marking. The timeline requires certain providers of AI systems, including general-purpose AI systems, that generate synthetic content and were already placed on the market before 2 August 2026, to comply with Article 50(2) by 2 December 2026. [5] The vendor resource records the deadline as a new transitional paragraph, Article 111(4). [6]
The amendment also simplifies or softens some obligations. [1] [6] The Commission lists a simplified AI literacy requirement, with the Commission and the Member States taking a stronger role in promoting literacy, an extension of measures previously reserved for SMEs to small mid-cap companies, expanded access to regulatory sandboxes with a new EU-level regulatory sandbox, and a simplified registration duty for exempted systems in the EU database. [1] The vendor resource reads the literacy change as shifting compliance from the level of literacy achieved to the measures taken, with the duty still covering every operator since 2 February 2025. [6]
How AI Office enforcement changes
The amendment widens the European AI Office's oversight, and the official pages describe its powers in operational terms. The office can request corrective action from providers of general-purpose AI models, restrict the public availability of a model in case of infringement, and issue direct fines for non-compliance. [3] Its investigations of general-purpose AI model providers range from inspections to requests for technical documentation and access to models. [3] The office employs more than 125 staff across 6 units. [3]
Under the amendments, the office has extended oversight of certain AI systems, including those built on general-purpose models and embedded in large online platforms and search engines. [3] [1] The vendor resource attributes the new enforcement chapter to Articles 75a to 75d and adds a mechanism the Commission pages do not state, periodic penalty payments of up to 5 percent of average daily turnover for each day a breach continues, which it notes can exceed the headline maximum on a long-running infringement. [6]
What enterprise teams can check before year end
Three verifiable items to close before year end. First, inventory generative capabilities against the two new prohibitions above. Second, confirm the machine-readable marking required by Article 50(2) on synthetic-content systems already placed on the market before 2 August 2026, a transitional deadline that binds certain providers on 2 December 2026. [5] [6] Third, re-check any compliance guidance dated before 27 July 2026, since the vendor resource warns that deadlines, prohibition counts and penalty mechanics in older material are likely to be wrong. [6]
Taken together, the amendment moved the high-risk conformity programs into December 2027 and August 2028 and front-loaded two prohibitions and one marking deadline into the remaining weeks of 2026. A program that targeted August 2026 now has until December 2027. The 2027 planning baseline is the Service Desk timeline, and the December work is machine-readable marking and the expanded prohibition list. [5] The dates above describe the regulation. They are not legal advice on any individual system.