Digital Omnibus on AI: Adjusted timelines and transitional provisions explained
As highlighted in our recent Law-Now, the European Commission took a significant step on 19 November 2025 by publishing the Digital Omnibus on AI (the “Omnibus”) (Digital Omnibus on AI Regulation Proposal | Shaping Europe’s digital future) – a legislative package aimed at refining, simplifying, and realigning selected provisions of the AI Act.
As part of our AI Watch series, we will be looking at each of the key parts of the Omnibus, and what these mean in practice.
First up: a closer look at the adjusted timelines and transitional provisions, and why they matter for your compliance strategy.
What the changes in timing are aiming to achieve:
At its core, the Omnibus aims to harmonise deadlines and streamline overlapping obligations, ensuring they follow a logical sequence.
High-risk systems:
This is particularly evident in the timing of obligations for high-risk systems. For example, the Omnibus connects entry into application of the high-risk requirements to the availability of harmonised standards, common specifications, and Commission guidelines.
Once the Commission confirms availability of standards and other support tools, high-risk rules will apply 6 months later for Annex III systems, and 12 months later for Annex I systems, subject to the following backstop dates:
- Annex III high-risk systems: latest by 2 December 2027
- Annex I high-risk systems: latest by 2 August 2028.
Watermarking for synthetic content:
A similar transitional regime applies to watermarking obligations under Art. 50 of the AI Act (covering synthetic audio, video, image and text content), whereby watermarking involves embedding a signal into an AI model to identify the content as AI generated. Providers of generative AI systems placed on the market before 2 August 2026 have until 2 February 2027 to integrate technically robust marking tools, benefitting from a six-month grace period.
Systems placed on the market after that date will not have the benefit of a general transitional period.
Comment:
These measures do not reduce obligations or change core safeguards. Instead, they acknowledge that compliance must be technically and institutionally feasible. This revised approach lowers execution risk while preserving the AI Act’s protective intent by synchronising application dates, introducing targeted transitional windows and calibrating enforcement.
Further updates on other key elements of the Omnibus will follow this piece.
