Governing AI-Generated Content: Labelling, Traceability and Transparency Infrastructure in China and the EU AI Act

Abstract This article examines how AI-generated content labelling is institutionalised as a legal-technical infrastructure for digital law and governance in China and the European Union. Focusing on China’s 2025 Labelling Measures, GB 45438-2025 and Article 50 of the EU AI Act, it asks how transparency obligations are translated into technical marking, metadata, platform verification and responsibility allocation. Methodologically, the article adopts comparative legal-policy analysis and structured document analysis of primary legal, regulatory, policy and technical-standard materials. It argues that AI-generated content labelling is neither a simple information-disclosure duty nor a purely technical detection problem. Rather, it functions as a compliance infrastructure connecting legal rules, technical standards, platform responsibility and user perception. The comparison shows that the EU model is primarily disclosure-oriented, relying on machine-readable marking, deepfake disclosure and provider/deployer accountability, whereas the Chinese model is more traceability-oriented, combining visible labels, implicit metadata, platform verification, application distribution review and user declaration. The article contributes to digital law and governance scholarship by showing how similar transparency vocabularies can generate different legal-technical infrastructures across jurisdictions.