On March 7, 2025, the Cyberspace Administration of China (CAC), the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration of Radio and Television jointly issued the Measures for the Labelling of Artificial Intelligence Generated Content (Measures) to mandate the labelling of AI-generated content (AIGC). The mandatory national standard Cybersecurity Technology - Methods for Identifying Artificial Intelligence Generated Content (GB 45438-2025) and the supporting practice guide Practice Guide for Cybersecurity Standards - Coding Rules for Labelling of Artificial Intelligence Generated Content Service Providers were also released simultaneously.
The Measures and the supporting mandatory national standards will take effect on September 1, 2025, giving relevant entities about six months to prepare for compliance.
Labelling AIGC with Explicit and Implicit Labels
The Measures require network information service providers to label AIGC with both explicit and implicit labels.
Explicit Labels: Explicit labels are added to AIGC or in interactive scenes in the form of text, sound, or image and can be easily perceived by users.
Service providers should add explicit labels to AI-generated text, images, audio, video, and virtual content that may cause confusion or misidentification according to the following requirements (Art.4)
- Text: Add text or general symbol labels at appropriate positions at the beginning, end, or middle of the text, or add prominent labels at the interface of interactive scenes or around the text.
- Audio: Add voice or audio rhythm labels at appropriate positions at the beginning, end, or middle of the audio, or add prominent labels at the interface of interactive scenes.
- Images: Add prominent labels at proper positions on images.
- Videos: Add prominent labels at appropriate positions on the starting screen of the video and the periphery of the video and add prominent labels at appropriate positions at the end and in the middle of the video.
- Virtual Scenes: Add prominent labels at appropriate positions on the starting screen and during the continuous service of the virtual scene.
- Other Scenarios: Add prominent labels according to their own application characteristics.
Service providers should also ensure that documents contain explicit labels that meet the requirements when providing functions for downloading, copying, and exporting AIGC.
Implicit Labels: Implicit labels are added by adopting technical measures in the file's data of AIGC and are not easily perceived by users.
Service providers must add implicit labels to the file's metadata of AIGC without impacting users' use of AIGC. These labels should include information on the nature of the AIGC, the name or code of the service provider, and content numbers, among other elements (Art. 5).
"File's metadata" refers to descriptive information embedded in the file header in a specific encoding format to record information including the source, nature, and purpose of the file.
Obligations of Service Providers
All service providers must provide methods and examples for labelling AIGC in the user service agreement and remind users to carefully read and understand the labelling requirements. If requested by the user, the service provider may provide AIGC without an explicit label, provided that the user's labelling obligations and liabilities are specified in the user service agreement, and the service provider must keep logs of the recipient’s information and other relevant information for no less than six months (Arts. 8 & 9).
When service providers conduct relevant procedures such as algorithm filings and security assessments, they should provide documents related to the labels of AIGC, strengthen the sharing of labelling information, and provide support in combating related illegal and criminal activities (Art. 12).
Obligations of Service Providers Offering Network Information Content Dissemination Services
Service providers that offer online content dissemination services are required to verify AIGC and ensure that appropriate labels are affixed, specifically (Art. 6):
- If the file's metadata has AI labels, add labels to inform the public that the content is synthetic.
- If the file's metadata lacks AI labels, but the user declares that the content is AIGC, add labels to inform the public that the content may be synthetic.
- If the file's metadata lacks AI labels and the user does not declare the AIGC, but the service provider detects labels or other traces of AIGC, add labels to inform the public that the content is suspected to be synthetic.
In these cases, information on the nature of AIGC, the name or code of the dissemination platform, content number, and other dissemination information should be added to the file's metadata. Network information content dissemination service providers should also provide necessary labelling functions and remind users to declare the AIGC.
App Distribution Platforms
When app distribution platforms conduct app listings or online reviews, they must require app service providers to disclose if they offer AIGC services and verify the related labelling materials (Art. 7).
Obligations of Users
Users are required to declare the AIGC and use the labelling function provided by the service provider when publishing such content (Art. 10).
Prohibited Acts
The following behaviors are strictly prohibited for any organization or individual (Art. 10):
- Malicious removal, alteration, or concealment of the AIGC labels.
- Providing tools or services for others to carry out the above malicious acts.
- Using improper labels to infringe on the legitimate rights and interests of others.
Liability
The Measures do not specify liability for violations but stipulate that the relevant competent authorities shall handle the matter in accordance with their duties and relevant laws and regulations.
Observations
The rise of domestically developed AI models such as DeepSeek has intensified regulatory discussions. This surge in AI development has highlighted the need for clear regulations to manage AIGC effectively. In response to this need, AI regulation has become a key focus of the CAC’s 2025 Qinglang, or "clear and bright," campaign – an annual initiative aimed at cleaning up China’s online activities. One of the key objectives this year is related to AI technology and AIGC, including strengthening the labelling requirements and combating the use of AI to generate and disseminate false information or to implement internet water armies, and regulating the ecosystem of AI applications.
Therefore, enterprises should align their practices with the standards and guidelines and take relevant measures to avoid potential penalties.
Key contacts
Tracy Chen
Associate , Shanghai (Kewei)
Disclaimer
The articles published on this website, current at the dates of publication set out above, are for reference purposes only. They do not constitute legal advice and should not be relied upon as such. Specific legal advice about your specific circumstances should always be sought separately before taking any action.