LEGAL NEWS IN THE SPOTLIGHT | EU Commission (AI Office) publishes the Code of Practice on Transparency of AI-Generated Content

On 10 June 2026, the European Commission, through the AI Office, published the Code of Practice on Transparency of AI-Generated Content (the Code), providing practical guidance on the implementation of the transparency obligations laid down in Article 50 of Regulation (EU) 2024/1689 (the AI Act). The Code was developed by independent experts through a multi-stakeholder process coordinated by the AI Office following the public consultation launched in September 2025. It will now be assessed by the European Commission and the AI Board before being complemented by the Commission’s forthcoming Guidelines on Article 50 AI Act.

The Code provides a practical compliance framework for the AI Act’s transparency obligations, distinguishing between two complementary sets of measures (a) those applicable to providers of generative AI systems, and (b) those applicable to deployers. For providers, it focuses on technical solutions ensuring that AI-generated content can be reliably identified through metadata, watermarking and detection mechanisms. For deployers, it sets out guidance on the disclosure of deepfakes and certain AI-generated text. Throughout the document, particular emphasis is placed on ensuring that transparency measures remain effective, interoperable and proportionate.

The Code also reflects a clear convergence between AI governance and data protection. In particular, providers offering detection services are expected to embed GDPR principles into the design of those services by implementing data minimisation, purpose limitation, storage limitation and appropriate security safeguards, while favouring privacy-preserving technologies and, where technically feasible, zero-retention processing of uploaded content.

Although compliance with the Code remains voluntary, the transparency obligations under Article 50 AI Act will become mandatory from 2 August 2026 (subject to the transitional regime applicable to certain AI systems already placed on the market). Once endorsed by the Commission, the Code is expected to become the principal benchmark for demonstrating compliance with those obligations, offering organisations greater legal certainty while reducing the need to justify alternative compliance measures on an individual basis.

Newsletter n. 120 – July 2026