The Measures create end-to-end labeling duties for generation services, content-distribution platforms, application stores and users. They entered into force on 1 September 2025 and are administered by the cyber, telecommunications, public-security and broadcasting authorities.
Regulation
Verified record
- Risk classification summary
- The Measures regulate generated and synthetic text, images, audio, video and virtual scenes through explicit and metadata-based implicit labels rather than risk classes.
- Requirements summary
- Generation providers must add prescribed visible labels and metadata. Distribution services must verify or detect labels and warn the public, application stores must verify labeling materials, and users publishing generated content must declare and label it.
- Affected entities
- AI generation and synthesis service providers, network content-distribution services, application-distribution platforms, and users publishing AI-generated or synthetic content.
- Enforcement summary
- Cybersecurity, telecommunications, public-security and radio-and-television authorities enforce the Measures within their respective responsibilities under Article 13 and related law.
- Penalty summary
- Violations are handled under applicable laws, administrative regulations and departmental rules; the Measures do not state one standalone universal fine schedule.
Source-backed duties
Verified legal obligations
Each duty below is tied to an official legal source and identifies who must act.
Provision: Articles 4–5
Add explicit and metadata-based labels at generation
AI generation and synthesis service providers must add prescribed explicit labels to generated or synthetic text, images, audio, video and virtual scenes, and must add implicit labels in file metadata.
- Applies to
- AI generation and synthesis service providers
Provision: Article 6
Verify labels and warn content recipients
Network content-distribution services must verify explicit labels or detect implicit labels, add the required notice to content, and warn the public when content may be AI-generated or synthetic.
- Applies to
- Network content-distribution services
Provision: Article 7
Verify labeling materials before application distribution
Application-distribution platforms must verify the AI-generated content labeling materials submitted by providers before applications are launched or listed.
- Applies to
- Application-distribution platforms
Provision: Articles 9–10
Declare generated content, preserve labels, and retain records
When providers supply content without an explicit label at a user’s request, they must agree the user’s labeling duty and retain relevant logs for at least six months. Users publishing AI-generated or synthetic content must declare it, add required labels, and must not maliciously delete, alter, forge, or conceal labels.
- Applies to
- AI generation and synthesis service providers
- Users publishing AI-generated or synthetic content