On March 11, 2026, the European Parliament and the Council reached a political agreement on the "Digital Omnibus on AI" — a text that modifies several deadlines under the EU AI Act (Regulation 2024/1689). Here is what this agreement actually changes for your business — and why starting your compliance journey now remains the most strategic decision.
What is the Digital Omnibus on AI?
Proposed by the European Commission on November 19, 2025, the Digital Omnibus is a legislative simplification package covering amendments to the GDPR, NIS2, Data Act, and EU AI Act. The goal: reduce administrative burden without weakening the regulation's fundamental protections.
On March 11, 2026, the Parliament and Council reached a political agreement on the AI Act amendments. The text must still be formally adopted and published in the EU Official Journal to take legal effect.
Key changes for SMEs
Delayed obligations for high-risk AI systems
The most widely covered measure. Obligations for high-risk AI systems (Annex III), originally due August 2, 2026, are proposed to be delayed to December 2, 2027 at the latest. Annex I systems (regulated products: medical devices, industrial machinery) would have until August 2, 2028.
- These are backstop dates, not automatic guarantees.
- Entry into application is linked to the availability of harmonised compliance standards, confirmed by a Commission decision.
- Obligations may apply earlier than the announced dates.
Extended simplified regime for mid-sized companies
The simplified EU AI Act regime, previously limited to SMEs (under 250 employees, under €50M revenue), is extended to mid-sized companies with up to 750 employees and €150M revenue: simplified compliance procedures, lighter documentation requirements.
European regulatory sandboxes
The agreement introduces an EU-level regulatory sandbox alongside existing national sandboxes. SMEs and mid-sized companies can pilot their AI solutions in real-world conditions under regulatory supervision before commercial deployment.
What does NOT change
- Absolute prohibitions (Title II) have been in force since February 2, 2025: psychological manipulation, social scoring, real-time facial recognition in public spaces. No changes.
- Transparency obligations (Article 50) remain fully applicable from August 2, 2026. Chatbots, AI-generated content, user-facing AI systems: disclosure is mandatory.
- GPAI obligations (Chapter V) have been in force since August 2, 2025.
5 reasons to act now despite the delay
-
The political agreement is not yet law. If the Digital Omnibus is not formally adopted before August 2, 2026, the original EU AI Act applies in full — no delay. Building compliance on a legislative bet is a real business risk.
-
Classifying your AI systems is unavoidable. Delay or not, you need to know which AI systems you use and their risk category. That's step zero — and most SMEs haven't done it yet.
-
Your clients and partners already require it. Large companies subject to earlier compliance timelines are beginning to require documentation from their suppliers. A classification attestation is an immediate competitive advantage.
-
Early compliance costs less. Acting under deadline pressure costs significantly more. Acting now gives you time to document, train your teams, and adjust processes without urgency.
-
It's a differentiation opportunity. In a market where AI trust is becoming a purchasing criterion, a proactive compliance approach sets you apart from competitors who are waiting.
Summary: partial delay, full risk
The Digital Omnibus brings real simplifications, particularly for high-risk systems and mid-sized companies. But it doesn't erase the EU AI Act — it adapts it. And it is not yet in force.
- Take inventory of your AI systems (5 minutes with our free tool)
- Identify priority obligations: Article 50 transparency → August 2, 2026 (firm) | high-risk → likely delay but uncertain
- Start documentation — even a delay doesn't justify inaction