Blog

What 2 December 2026 means under the AI Act, in plain language

On 2 December 2026 a four-month transition ends for marking AI-generated content, and two new prohibitions start to apply. Who is affected, who is not, and what to check before the date.

4 min

Two things happen under the AI Act on 2 December 2026. One is the end of a short transition for marking AI-generated content. The other is the start of two new prohibitions. Neither is a new wave of obligations for every company, and neither is a postponement of anything. Here is what the date means, with the legal text next to each point.

The short version

  • If you build or sell an AI system that generates text, images, audio or video, and it was already on the market before 2 August 2026, you have until 2 December 2026 to make its output machine-readably marked and detectable.
  • If you only use AI tools, this deadline is not addressed to you. Your own transparency duties have applied since 2 August 2026 and do not change on 2 December.
  • Two new prohibitions, added to the AI Act this summer, apply from the same day.

1. The transition for marking AI-generated content ends

Article 50(2) of the AI Act requires providers of generative AI systems to make sure the content their systems produce is marked in a machine-readable format and detectable as artificially generated. Article 50 has applied since 2 August 2026.

The Digital Omnibus on AI (Regulation (EU) 2026/1744) added one transition, in the new Article 111(4). Providers of systems "placed on the market before 2 August 2026" must "take the necessary steps in order to comply with Article 50(2) by 2 December 2026".

Three limits are worth reading closely:

  • It covers providers only. A provider is the company that develops the system or puts it on the market under its own name. Deployers, the companies that use a system in their work, are not the addressees of Article 50(2).
  • It covers older systems only. A generative system first placed on the market on or after 2 August 2026 had to comply from its first day.
  • It covers marking and detection only. The European Commission describes it as "a limited grace period", "only as regards the marking and detection obligation".

2. Two new prohibitions start to apply

The same regulation added new prohibited practices to Article 5 of the AI Act. Under the amended Article 113, they "shall apply from 2 December 2026". The Commission's AI Act timeline describes them as prohibitions on AI systems that generate non-consensual sexual deepfakes and child sexual abuse material.

For almost every company this changes nothing in daily work. It matters if you build or offer image or video generation: check that your product cannot be used for these purposes and that your terms of use say so.

What does not change on 2 December

  • Telling people they are talking to an AI (Article 50(1)): applies since 2 August 2026, with no transition. The Commission's guidelines say a system that both chats and generates content gets the extra time for marking only, not for this disclosure.
  • Informing people about emotion recognition or biometric categorisation (Article 50(3)): applies since 2 August 2026.
  • Labelling deepfakes and AI-generated text on matters of public interest (Article 50(4)): applies since 2 August 2026. This one is addressed to deployers.
  • High-risk rules: still 2 December 2027 for the Annex III systems and 2 August 2028 for AI embedded in regulated products. The year is different. The two December dates are easy to confuse.
  • Older content: the Commission states that "content generated prior to 2 August 2026 does not need to be labelled retroactively".

What to check before 2 December

If you provide a generative AI system:

  1. Confirm when it was first placed on the EU market. That date decides whether the transition applies to you at all.
  2. Check that its output carries a machine-readable mark and can be detected as AI-generated. The Commission's Code of Practice on marking and labelling, published on 10 June 2026, is a voluntary way to show this.
  3. Write down what you implemented and when.

If you use generative AI tools from someone else:

  1. Ask your vendor, in writing, whether their output is marked in line with Article 50(2) and from which date.
  2. Review your own duties, which already apply: chatbot disclosure, deepfake labelling, labelling of AI-written text on matters of public interest that nobody reviewed editorially.
  3. Keep the vendor's answer with your AI inventory.

Our platform keeps that inventory, the obligation plan and the evidence in one place, including the Article 50 workflow.

What we could not confirm

The Commission published its Article 50 guidelines on 20 July 2026 and stated that formal adoption would follow once all language versions were available. We have not found a notice that this step has happened. The guidelines are not binding in any case. The regulation is, and the quotes above come from its text.

Sources

Ready to be compliant?

Start free today. 14 days, no credit card required.