China's AI-Generated Content Labelling Measures
China’s labelling rules are the most prescriptive synthetic-content regime currently in force anywhere. They require two labels rather than one — a visible mark for people and a metadata mark for machines — and they distribute the obligation across generators, distribution platforms, app stores and users, so no single party can discharge it alone. The instrument and its date The Measures for…
China has implemented the strictest labeling rules for AI-generated content in the world. These measures require two labels - one visible for users and one hidden in metadata for machines - to be applied by generators, platforms, app stores, and users. Issued on March 14, 2025, the Measures for Labelling of AI-Generated Synthetic Content came into effect on September 1, 2025.
The rules are outlined in a national standard (GB 45438-2025) which specifies where and how to encode the labels. The labeling system builds upon three previous instruments dealing with deep synthesis content and generative AI services. The labeling requirements aim to prevent confusion when the synthetic nature of content could mislead users.
The explicit label must be visible in text, image, video, audio, virtual scenes, and interactive content. The implicit label consists of metadata embedded in the file format that identifies the content as AI-generated, the service provider's name or code, and a content identification number. The four-party duty chain involves generators, platforms, app stores, and users all sharing the responsibility of applying and detecting labels.
Violations can be enforced through existing supervisory authorities. This labeling regime differs from the EU's approach which requires machine-readable marking from providers and separate human-facing disclosure.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
