The European Commission published the final version of the Code of Practice on Marking and Labelling of AI-Generated Content on 10 June 2026, as provided for by Article 50 of the AI Act.
The document represents the outcome of eight months of work, including dozens of meetings, workshops, and bilateral discussions involving more than 200 stakeholders and representatives of all 27 European Union Member States.
Two working groups, composed of experts in technology, law, and the social sciences, contributed to defining a framework aimed at preventing deception and manipulation associated with synthetic content and at strengthening public trust in the digital ecosystem.
Professor Dino Pedreschi served as Vice-Chair of Working Group 1, Transparency Obligations Applicable to Providers of Generative AI Systems, which was responsible for developing guidelines on transparency and the identification of content generated by generative artificial intelligence systems. He was supported in this work by Virginia Morini, a postdoctoral researcher at the Department of Computer Science, and Daniele Gambetta, a PhD Student in AI.
“Coordinating this work was both a challenging and fascinating experience. We now hope that many stakeholders will adopt the Code and that it will provide an effective contribution to the fight against misinformation and disinformation by making the human or artificial origin of online content increasingly recognizable. An important aspect is that this work has already inspired similar legislative initiatives in several parts of the world, particularly in California, where we are collaborating with local lawmakers,” said Dino Pedreschi.
Professor Pedreschi also noted that, on 11 June, OpenAI announced its support for the initiative, describing it as a useful instrument for fostering an innovative and trustworthy European AI ecosystem.
Further information is available through the official communication channels of the European Union.