The Cyberspace Administration of China has released the Draft Measures for the Identification of AI-Generated Content, which is formulated in accordance with the Cybersecurity Law of the People's Republic of China, the Regulations on the Management of Algorithmic Recommendations in Internet Information Services, the Regulations on the Management of Deep Synthesis in Internet Information Services, and the Interim Measures for the Management of Generative Artificial Intelligence Services, among other laws and regulations. The deadline for feedback is October 14, 2024.

The Measures define AI-generated content to include text, images, audio, video, and other information, distinguishing between explicit and implicit identifiers. Explicit identifiers are prominently displayed in the content or user interface, while implicit identifiers are added through technical means within the content file data, which are not easily perceptible to users.

The Measures stipulate that service providers should add explicit identifiers at appropriate locations according to the type of content when providing generative synthesis services, and add implicit identifiers in the file metadata. Additionally, they require online information content dissemination platform providers to take measures to regulate the dissemination of generative synthesis content, including verifying implicit identifiers in the file metadata and adding warning identifiers around the published content.

WeChat Screenshot_20240914110802.png

The following is the original text of the draft:

Article 1 These Measures are formulated to promote the healthy development of artificial intelligence, regulate the identification of AI-generated content, protect the legitimate rights and interests of citizens, legal entities, and other organizations, and maintain public interests, in accordance with the Cybersecurity Law of the People's Republic of China, the Regulations on the Management of Algorithmic Recommendations in Internet Information Services, the Regulations on the Management of Deep Synthesis in Internet Information Services, the Interim Measures for the Management of Generative Artificial Intelligence Services, and other laws, administrative regulations, and departmental rules.

Article 2 These Measures apply to network information service providers (hereinafter referred to as "service providers") who, in accordance with the Regulations on the Management of Algorithmic Recommendations in Internet Information Services, the Regulations on the Management of Deep Synthesis in Internet Information Services, and the Interim Measures for the Management of Generative Artificial Intelligence Services, carry out the identification of AI-generated content.

Industry organizations, enterprises, educational and research institutions, public cultural institutions, professional institutions, and others that develop and apply AI-generated synthesis technology but do not provide services to the public in China are not subject to these Measures.

Article 3 AI-generated content refers to information such as text, images, audio, and video produced, generated, or synthesized using artificial intelligence technology.

AI-generated content identification includes explicit and implicit identifiers.

Explicit identifiers are those added in the AI-generated content or interactive interface, presented in text, sound, graphics, and other forms, which can be clearly perceived by users.

Implicit identifiers are those added through technical means within the AI-generated content file data, which are not easily perceptible to users.

Article 4 Service providers whose generative synthesis services fall under the circumstances described in the first paragraph of Article 17 of the Regulations on the Management of Deep Synthesis in Internet Information Services shall add explicit identifiers to the generative synthesis content in accordance with the following requirements.

(1) Add text or symbol prompts at the beginning, end, or appropriate locations within the text, or add prominent warning identifiers around the text in the interactive interface;

(2) Add voice or audio rhythm prompts at the beginning, end, or appropriate locations within the audio, or add prominent warning identifiers in the interactive interface;

(3) Add prominent warning identifiers at appropriate locations within the image;

(4) Add prominent warning identifiers at appropriate locations within the video start screen and around the video playback, and may add prominent warning identifiers at appropriate locations within the video;

(5) When presenting virtual scenes, add prominent warning identifiers at appropriate locations within the start screen, and may add prominent warning identifiers at appropriate locations during the continuous service of the virtual scene;

(6) Other generative synthesis service scenarios should add explicit identifiers with significant warning effects based on their application characteristics.

When service providers offer generative synthesis content for download, copying, exporting, and other methods, they should ensure that the files contain explicit identifiers that meet the requirements.

Article 5 Service providers should add implicit identifiers in the file metadata of generative synthesis content in accordance with Article 16 of the Regulations on the Management of Deep Synthesis in Internet Information Services. Implicit identifiers should include attribute information of generative synthesis content, the name or code of the service provider, content number, and other production element information.

Service providers are encouraged to add digital watermarks and other forms of implicit identifiers in generative synthesis content.

File metadata refers to descriptive information embedded in the file header in a specific encoding format, used to record file source, attributes, purpose, copyright, and other information.

Article 6 Service providers offering online information content dissemination platform services should take measures to regulate the dissemination activities of generative synthesis content.

(1) Verify whether implicit identifiers are included in the file metadata. For those containing implicit identifiers, take appropriate measures to add prominent warning identifiers around the published content, clearly reminding users that the content is generative synthesis content;

(2) If no implicit identifiers are verified in the file metadata but the user declares it as generative synthesis content, take appropriate measures to add prominent warning identifiers around the published content, reminding users that the content may be generative synthesis content;

(3) If no implicit identifiers are verified in the file metadata, the user has not declared it as generative synthesis content, but the service provider offering online information content dissemination platform services detects explicit identifiers or other generative synthesis traces, it can be identified as suspected generative synthesis content. Appropriate measures should be taken to add prominent warning identifiers around the published content, reminding users that the content is suspected of being generative synthesis content;

(4) For content that is, may be, or is suspected of being generative synthesis content, add attribute information of generative synthesis content, the name or code of the dissemination platform, content number, and other dissemination element information in the file metadata;

(5) Provide necessary identification functions and remind users to actively declare whether the published content includes generative synthesis content.

Article 7 Internet application distribution platforms should verify whether service providers have provided the required generative synthesis content identification functions during application listing or online review.

Article 8 Service providers should clearly explain the methods and styles of generative synthesis content identification in the user service agreement, and prompt users to carefully read and understand the relevant identification management requirements.

Article 9 If users require service providers to provide generative synthesis content without explicit identifiers, they may provide generative synthesis content without explicit identifiers after clarifying the user's identification obligations and usage responsibilities through the user agreement, and retain relevant logs for no less than six months.

Article 10 When users upload generative synthesis content to service providers offering online information content dissemination platform services, they should actively declare and use the platform's identification functions to identify the content.

No organization or individual may maliciously delete, alter, forge, or conceal the generative synthesis content identifiers as stipulated by these Measures, provide tools or services for such malicious acts, or damage the legitimate rights and interests of others through improper identification means.

Article 11 Service providers should conduct identification in accordance with the relevant mandatory national standards.

Article 12 Service providers should provide relevant materials on generative synthesis content identification in accordance with these Measures when performing algorithm filing, security assessment, and other procedures, and strengthen the sharing of identification information to provide support and assistance for preventing and combating related criminal activities.

Article 13 Those who violate these Measures by failing to identify generative synthesis content and cause serious consequences will be punished by the relevant authorities such as the Cyberspace Administration in accordance with the relevant laws, administrative regulations, and departmental rules.