On 10 June 2026, the European Commission published the final Code of Practice on Transparency of AI-Generated Content, completing a drafting process initiated in 2025 under Article 56 of Regulation (EU) 2024/1689. The Code operationalises mandatory transparency obligations under Articles 50(2), 50(4), and 50(5) of the EU AI Act, which require providers and deployers of certain AI systems to ensure that AI-generated audio, video, image, and text content is machine-detectable and, in defined contexts, visibly labelled. These obligations apply from 2 August 2026.
The Code is structured in two sections. Section 1 sets technical standards for machine-readable marking of AI-generated content applicable to providers of generative AI systems, covering watermarking, cryptographic provenance metadata, and detection interoperability. Section 2 sets visible labelling commitments for AI-generated or AI-manipulated deepfake audio and video under Article 50(4), and for AI-generated or AI-manipulated text published to inform the public on matters of public interest under Article 50(5). Providers and deployers that sign the Code and adhere to its requirements can rely on it as a compliance pathway under Article 50.
Providers of general-purpose AI models with generative output capabilities must implement content marking at the model or output layer and make detection capabilities accessible to downstream deployers. Broadcasting organisations, online platforms, and firms deploying generative AI to produce news content, advertising, or public-interest communications fall within deployer obligations under Article 50(4) and (5). Compliance requires both technical implementation and governance documentation capable of demonstrating conformity to the AI Board and national competent authorities.
The Code does not extend Article 50 obligations beyond the AI Act's own scope, and it does not apply to content produced for research, national security, or law enforcement purposes. AI systems generating content in response to explicit human creative direction may claim reduced labelling obligations under Article 50(4)(b). The Commission and AI Board are assessing whether current signatories and commitments are adequate; if found insufficient, the Commission may issue implementing acts to impose binding transparency measures on non-signatories.
Licentium assists AI system providers, deployers, and media organisations with EU AI Act compliance, including Article 50 obligation mapping and content governance design. We may advise on this matter or refer within our partner network. Work we undertake includes EU AI Act Article 50 transparency compliance, Code of Practice signatory assessment, general-purpose AI model regulatory analysis, AI content labelling implementation review, and AI governance documentation.
Source: European Commission, Code of Practice on Transparency of AI-Generated Content, 10 June 2026