This morning session brings together officials from the European AI Office, academia, industry and civil society to discuss the new Code of Practice.
Registration is possible both online and on premise.
Article 50 of the AI Act has applied since 2 August 2026, requiring providers of generative AI systems to make outputs machine-readable as AI-generated, and deployers to disclose deepfakes and certain AI-generated text. To operationalise these duties, the AI Office coordinated a multi-stakeholder drafting process led by independent chairs across two working groups, marking and detection for providers, and disclosure obligations for deployers. The resulting Code of Practice was published in June 2026, assessed as adequate by the Commission and the AI Board, and signed by roughly 190 organisations.
This morning session brings together the AI Office, chairs and vice-chairs of both working groups, and a small group of academics, industry and civil-society participants to take stock.
Discussion will focus on three questions: whether watermarking, digitally signed provenance metadata and the common EU label hold up in real deployment; how to label AI-generated and modified content; and how adherence will be monitored, articulated with the DSA and international provenance standards.
AGENDA
10:00–10:10
Opening remarks and introduction
10:10–10:35
Keynote speech - European Commission, AI Office
10:35–11:20
Session I — Section 1: marking and detection for providers
11:20–11:35
Coffee break
11:35–12:20
Session II — Section 2: labelling deepfakes and AI-generated text for deployers
12:20–12:50
Academic and Policy Roundtable
12:50–13:00
Closing remarks
Register