On 19 November 2025 the European Commission proposed a Digital Omnibus on AI, amending the AI Act and related rules in the civil aviation field.
What the proposal tries to fix
The EU AI Act is the world’s first dedicated AI law built on risk tiering, but the complexity of its rules has drawn sustained industry criticism. The proposal’s starting point is to reduce compliance burdens — especially for SMEs — without abandoning the risk-tiering framework.
From the published content, the adjustments concentrate on three areas: simplifying the implementation rules for high-risk AI systems; expanding testing and experimentation (sandbox) opportunities; and adjusting the application timetable for certain obligations.
The most immediate effect is the timetable — the application dates for high-risk AI systems are postponed, with a further postponement for AI embedded in products. For companies this is both a buffer and a warning: “wait for the rules to settle” is no longer a safe strategy, because the obligations are postponed, not cancelled.
Three implications for companies
First, re-plan the compliance timetable. Product certification, documentation and conformity assessment originally scheduled against the earlier dates can shift somewhat; but technical documentation, data governance and risk management are “early is better” work and should not be shelved because of the delay.
Second, the value of sandbox participation rises. Expanded testing opportunities make validating high-risk functions in a controlled environment more feasible. Participation usually carries filing and disclosure duties, so companies should weigh the benefit against the disclosure cost.
Third, transparency duties are unaffected. It is important to distinguish “postponed high-risk duties” from “transparency duties already in force”. Transparency requirements for general-purpose AI models and marking of deepfake content apply on the original timetable. See our note on the EU AI Act transparency rules taking effect.
The legislative process ahead
The proposal needs to be considered by the European Parliament and the Council. In March 2026 the Parliament adopted its position, supporting a postponement of certain high-risk rules and a ban on “nudifier” apps. See our notes on the Parliament’s position and the nudifier ban.
For companies going global, the pace of change in EU rules warrants continued tracking. Treating “rules may change” as normal planning beats betting on a single point in time. See our EU AI Act service.
This article is based on the European Commission’s Digital Omnibus on AI proposal and public information. It is for general reference only and does not constitute legal advice.