On Wednesday 17 December, the European Commission published a first draft of the Code of Practice on the transparency of AI-generated content, which is intended to provide model providers with details on compliance with the transparency rules laid down in Article 50 of the AI Regulation.
In practice, however, suppliers are free to decide whether or not to adhere to it, as the Code is primarily a general recommendation for compliance, not a binding text.
This first version of the Code details how companies can correctly report that AI-generated content, such as an image or video appearing to show a real person, is in fact media that has been manipulated and is also known as ‘deepfake’.
This first draft of the Code comprises two sections. The first covers rules for suppliers of generative AI systems, focusing on the marking of AI-generated content and technical solutions; the second, whose rules apply to system deployers, covers the disclosure of ‘deepfakes’ and AI-generated or manipulated content on issues of public interest.
The text issued by the Commission introduces the possibility of seeing the emergence of a “common icon” to identify deepfakes generated by AI. This icon could resemble what is already used by certain platforms, such as a two-letter acronym referring to artificial intelligence (AI, KI, IA).
The Commission will be collecting comments from stakeholders on the first draft until 23 January. The second draft will be written by mid-March 2026, and the Code should be finalised by June 2026. The rules on the transparency of AI-generated content will come into force on 2 August 2026.
Link to the Code: https://aeur.eu/f/k1r (Original version in French by Isalia Stieffatre)