Labelling AI-Generated Content in China
China has implemented regulations requiring labeling of AI-generated content to combat misinformation and protect intellectual property. The rules mandate clear disclosure of AI involvement in text, images, audio, and video content, with non-compliant platforms facing penalties. These measures reflect China's broader efforts to govern artificial intelligence technologies.
Background
China's 2022 "Deep Synthesis Provisions" and the 2023 "Measures for the Management of Generative AI Services" require that AI-generated text, images, audio, and video be clearly labelled — both embedded programmatically (e.g. metadata) and presented visibly to users. These regulations, issued by the Cyberspace Administration of China (CAC), are part of a broader push to control disinformation and assert state oversight over AI platforms. Major Chinese tech firms (Baidu, Alibaba, Tencent, ByteDance) must comply. The rules apply not just to deepfakes but to any synthetic content produced by generative AI models, and require companies to watermark outputs, keep training data logs, and prevent generation of illegal or politically sensitive content. This contrasts with the largely voluntary, industry-driven labelling frameworks in the US and EU (such as the C2PA standard and the EU AI Act's transparency provisions), where government mandates are less prescriptive and enforcement is more fragmented.