The European Commission has issued a second draft of its Code of Practice aimed at helping providers and deployers meet the marking and labelling requirements for AI generated content under Article 50 of the AI Act.
The second version has been streamlined and simplified, with the aim of providing more flexibility for the signatories, reducing the compliance burden and incorporating further technical considerations to improve legal clarity and practicality. It promotes the use of open standards for AI content marking and an EU icon for labelling to simplify compliance and reduce costs. It is structured into two sections, each dealing with different aspects of transparency and regulation for providers and deployers respectively.
Section 1
Section 1 addresses marking and detecting AI content and is aimed at providers of generative AI systems within the scope of Article 50(2) AI Act. This version removes and consolidates several measures, and introduces optional elements, while aiming to make sure that all measures remain technically feasible and proportionate. Key commitments include a revised two-layered marking approach involving secured metadata and watermarking, optional fingerprinting and logging, and protocols for detection and verification.
Section 2
Section 2 targets deployers of AI systems. It focuses on labelling deepfakes and text publications concerning matters of public interest within the scope of Article 50(4) AI Act. Relative to the first draft, this section adopts a more flexible and practice-oriented approach. Since the first draft, it has been restructured to simplify and streamline the commitments, while the taxonomy distinguishing the AI–generated content from AI-assisted content has been completely removed. It now features design and placement requirements applicable to icons, labels or disclaimers, ensuring a minimum level of uniformity, while enabling signatories to conceive solutions adapted to their needs. In addition, the section proposes a task force to develop a future, uniform, interactive EU icon, with discretionary support from signatories.
The code has also further defined the specific regimes applicable to artistic, creative, satirical and fictional works and text publications under human review or editorial control, facilitating reliance on existing practices or procedures.
The annex of the second draft now includes illustrative examples of a potential EU icon to be made available to signatories.
Next steps
The code is expected to be finalised by beginning of June this year. The rules covering the transparency of AI-generated content will become applicable on 2 August 2026.
