On 18 March 2026, the European Parliament’s IMCO (Internal Market and Consumer Protection) and LIBE (Civil Liberties, Justice and Home Affairs) committees adopted a joint position on the so-called “Digital Omnibus” package — a sweeping legislative amendment that touches the EU AI Act, the EU Data Act, and the Product Liability Directive in a single package. For compliance teams, this is critical reading: some high-risk AI deadlines could shift, while the watermarking timeline may actually accelerate.
- Some high-risk AI deadlines may be extended by up to 12 months for specific Annex III categories where technical standards are not yet finalised — particularly HR AI and credit scoring AI.
- Article 50 watermarking requirements may move earlier — the joint position proposes bringing forward the AI-generated content labeling obligation from August 2026 to Q1 2026 for large platforms.
- This is a committee position, not yet law. It must pass full Parliament plenary, then be negotiated with the Council before becoming binding. Timeline: Q3–Q4 2026 at earliest.
- Do not pause your compliance programme. The August 2026 deadline remains in force unless and until the Digital Omnibus passes in final form. Planning around a potential extension is a compliance risk.
What Is the Digital Omnibus Package?
The “Digital Omnibus” is an informal name for a legislative package that consolidates amendments to multiple digital laws into a single instrument — a common EU legislative technique for reducing fragmentation and ensuring coherence across related regulatory frameworks. This iteration bundles proposed amendments to the EU AI Act, the EU Data Act, and the revised Product Liability Directive.
The joint IMCO/LIBE committee position adopted on 18 March 2026 reflects two years of practical experience watching organisations struggle to implement the EU AI Act’s technical requirements — and recognises that in some sectors, the harmonised standards needed for conformity assessments have not been published by CEN-CENELEC in time for the August 2026 deadline to be practical.
At the same time, the committees responded to growing public and civil society pressure over the proliferation of AI-generated disinformation and non-consensual synthetic imagery — pushing for some transparency obligations to be brought forward rather than delayed.
Which Deadlines May Change — and Which May Accelerate?
The committee position proposes two types of deadline change — extensions for some high-risk obligations, and acceleration for some transparency obligations. Here is the current picture as understood from the adopted committee position:
| Obligation | Current Deadline | Proposed Change | Direction |
|---|---|---|---|
| High-risk AI — HR & recruitment (Annex III, Cat. 4) | 2 August 2026 | Proposed extension to August 2027 pending finalisation of harmonised standards | ⬅ Later |
| High-risk AI — credit scoring & insurance (Annex III, Cat. 5) | 2 August 2026 | Proposed extension to August 2027 where standards not published by June 2026 | ⬅ Later |
| High-risk AI — biometrics, law enforcement, migration (Annex III, Cat. 1, 6, 7) | 2 August 2026 | No change proposed — deadline remains | Unchanged |
| Article 50 — AI content labeling for large platforms (>45M EU users) | 2 August 2026 | Proposed acceleration to 1 May 2026 for platforms meeting DSA “Very Large” threshold | ➡ Earlier |
| Article 50 — AI content labeling for all other operators | 2 August 2026 | No change proposed — deadline remains | Unchanged |
| Prohibited practices (Article 5) — all categories | 2 February 2025 (already in force) | No change proposed — in force and binding | Already active |
The Watermarking Acceleration: What Large Platforms Face Now
The most immediately urgent element of the Digital Omnibus position — if it becomes law — is the proposed acceleration of the Article 50 AI content labeling obligation for Very Large Online Platforms (VLOPs) as defined under the Digital Services Act. The committee proposes that platforms with more than 45 million EU monthly active users must implement AI-generated content watermarking by 1 May 2026 — three months earlier than the current August 2026 deadline.
The rationale cited in the committee’s explanatory memorandum is the 2026 election cycle across multiple EU member states and the growing evidence of AI-generated disinformation in political campaigning — including AI-generated audio impersonations of political candidates and AI-generated video of protest events that did not occur.
Practically, this means large platforms are being asked to implement C2PA-standard machine-readable watermarking (or equivalent) across AI-generated image, video, and audio content within weeks rather than months — before the EU AI Office has even published its final technical specifications for the “EU AI Icon” (see our separate news post on the Transparency Code of Practice Draft 2).
The Coalition for Content Provenance and Authenticity (C2PA), which develops the open technical standard for content credentials, has confirmed that its specification is compatible with the committee’s proposed watermarking requirements — providing a ready-made implementation pathway for platforms that have not yet begun this work.
