Brief

EU AI Act 2 December 2026 Deadline: What Businesses Must Do

2 December 2026 is not when the whole EU AI Act starts. It is the Article 50(2) transition for older systems, plus new prohibitions.

European Commission, AI Act Service Desk FAQ

What happened

The European Commission’s AI Act Service Desk FAQ, opened on November 19, 2026, sets out what changes on 2 December 2026. Transparency obligations enter into application and are enforceable from 2 August 2026. Providers of systems placed on the market before that date have until 2 December 2026 to comply with Article 50(2) marking and detection. The FAQ also names new prohibitions that will apply from 2 December 2026. They cover generation or manipulation of non-consensual intimate material and child sexual abuse material.

The Commission’s implementation timeline matches that split. It describes 2 December 2026 as the date when those new prohibitions apply. It is also the transitional deadline for certain providers of systems that generate synthetic content. That includes general-purpose AI systems already on the market before 2 August 2026. High-risk rules for Annex III systems apply from 2 December 2027. Rules for high-risk AI embedded in regulated products apply from 2 August 2028. Those later dates follow the Digital Omnibus on AI, the timeline says.

What it means

2 December 2026 is not the day the entire AI Act starts. Most transparency rules, and enforcement for applicable rules, already started on 2 August 2026. The December date is a transition for Article 50(2) marking and detection on systems already on the market. It is also the start date for the new prohibitions named above. High-risk duties that cover many workplace and credit tools are still a 2027 problem, on the Commission’s own timeline.

If you placed a generative system on the EU market after 2 August 2026, the FAQ’s August transparency date already applied. The December transition described here is for providers of systems placed on the market before 2 August 2026.

Why it matters to you

If you provide an AI system already used in the EU, you must check Article 50(2) marking and detection before 2 December 2026. If you deploy models or tools from a vendor, you can ask for written confirmation of that marking. Keep the answer with your compliance file. The new prohibitions apply on the same day, so legal and risk teams must know what the company will refuse to build or buy. High-risk classification for Annex III systems, including many uses in employment and access to essential services, still has until 2 December 2027 on the timeline. That is the next planning date, not an excuse to ignore December.

What to watch

The date to watch first is 2 December 2026. After that, the Commission’s timeline points to 2 August 2027, when each Member State is to have at least one AI regulatory sandbox. Then 2 December 2027, when Annex III high-risk rules apply. Watch the Service Desk FAQ for updates, because the Commission says the list will be updated regularly. This brief relies on the Commission Service Desk pages opened on November 19, 2026.

Sources

  1. European Commission, AI Act Service Desk FAQ
  2. European Commission, Timeline for the Implementation of the EU AI Act