The EU shifts into reverse
On 19 November 2025, the European Commission proposes postponing the central obligations of its own AI Act — a year and a half before they were due to take effect.
The EU AI Act entered into force in August 2024. Its weightiest obligations — those for so-called high-risk systems — were due to apply from 2 August 2026. On 19 November 2025, the European Commission put forward a proposal to postpone precisely that deadline.
The paper is called the «Digital Omnibus on AI». The name says it all: an omnibus proposal bundles amendments to several legal acts into one package. The European Parliament's research service has documented the process: EPRS briefing: Digital Omnibus on AI.
What was to be postponed
- Stand-alone high-risk systems: from 2 August 2026 to 2 December 2027.
- AI in regulated products: from 2 August 2027 to 2 August 2028.
The reasoning is unspectacular and credible for exactly that reason: the technical standards companies were supposed to align themselves with were not finished. In several member states, the authorities meant to supervise the whole thing were also missing.
An obligation without a yardstick cannot be met — and cannot be checked.
Why this matters for Swiss businesses
The AI Act does not stop at the EU's external border. It also applies to providers outside the Union if the output of a system is used within the EU. Anyone operating an AI-supported service from Switzerland for customers in Germany or France is covered — regardless of where the server stands.
A proposal, however, is not yet a law. Between 19 November 2025 and the day the postponement actually applied lay eight months of legislative procedure — with one failed round of negotiations in between. How that turned out is set out in the articles on March, June and August 2026.
In practical terms that means: you do not replan on the strength of a proposal. You read it, note the date and see what becomes of it.
Sources
This is how our system does it