What do China’s AI content labeling rules require?

Since September 1, 2025, AI-generated or synthetic content served in mainland China has to be labeled, and most consumer AI features need two labels. Every output carries an implicit label in its file metadata that marks it as AI-generated and identifies the service provider (Article 5). Services that could mislead the public, such as chatbots, synthetic voices, face generation and immersive scenes, must also add an explicit label people can see or hear, placed according to the medium (Article 4). Platforms that publish user content must check uploads for these labels and flag content that is, may be, or is suspected to be AI-generated (Article 6). App stores must ask every app whether it provides AI generation or synthesis services and verify its labeling materials (Article 7), the rule behind Apple’s AIGC declaration for China mainland. Labels do not replace the CAC filing that decides whether a generative AI service may run in China at all.

The articles that change what you ship

FactPrimary source
Article 3 defines two labels: an explicit label on the content or its interface that users can clearly perceive as text, sound or graphics, and an implicit label added to the file data through technical measures that users cannot readily perceive. Measures for Labeling AI-Generated or Synthetic Content (CAC, MIIT, MPS, NRTA, 2025), Article 3
Article 4 places the explicit label by medium: a text or symbol prompt for text, a voice or rhythm prompt for audio, a conspicuous mark on images, the opening frame and the area around playback for video, and the opening screen of a virtual scene. Downloaded, copied or exported files must keep it. Measures for Labeling AI-Generated or Synthetic Content (CAC, MIIT, MPS, NRTA, 2025), Article 4
Article 5: the implicit label goes into the file metadata and records that the content is generated or synthetic, the service provider’s name or code, and a content number. Digital watermarks are encouraged. Measures for Labeling AI-Generated or Synthetic Content (CAC, MIIT, MPS, NRTA, 2025), Article 5
Article 7: app distribution platforms must ask, at listing or launch review, whether an app provides AI generation or synthesis services, and must verify its labeling materials if it does. Measures for Labeling AI-Generated or Synthetic Content (CAC, MIIT, MPS, NRTA, 2025), Article 7
Article 9: content without the explicit label may be supplied at a user’s request only after the user agreement sets out the user’s labeling obligations, and the provider must keep the related logs, including who received it, for at least six months. Measures for Labeling AI-Generated or Synthetic Content (CAC, MIIT, MPS, NRTA, 2025), Article 9

Sources verified by the 21YunBox compliance team on 2026-11-18.

TLDR; Below is the English-translated version of China’s Measures for Labeling AI-Generated or Synthetic Content (人工智能生成合成内容标识办法), issued March 7, 2025 by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration under notice Guoxinban Tongzi [2025] No. 2, and in effect since September 1, 2025.


If your product generates text, images, audio, video, or virtual scenes for users in mainland China, or lets users publish content that may be AI-generated, these Measures decide how that content has to be labeled. They require an explicit label that people can see or hear and an implicit label written into the file’s metadata, they make content platforms check for those labels and flag AI-generated posts, and in Article 7 they require app stores to ask every app whether it provides AI generation or synthesis services, then verify the labeling materials of those that do. That article is the rule behind the AIGC question App Store Connect now asks for China mainland.

The Measures put into practice the labeling duties already set by the Deep Synthesis Provisions (Articles 16 and 17), the Generative AI Measures, and the Algorithm Recommendation Provisions, and they apply alongside the mandatory national standard GB 45438-2025, which took effect the same day. Article 12 also asks for labeling materials when a provider completes algorithm filing or a security assessment. Correct labels are a condition of running an AI feature in China, not a replacement for the filing that decides whether the feature may run there at all; confirm both with counsel.


Measures for Labeling AI-Generated or Synthetic Content (Issued March 7, 2025 by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration; effective September 1, 2025)

Article 1. These Measures are formulated in accordance with the “Cybersecurity Law of the People’s Republic of China,” the “Provisions on the Administration of Algorithm Recommendation for Internet Information Services,” the “Provisions on the Administration of Deep Synthesis of Internet Information Services,” the “Interim Measures for the Management of Generative Artificial Intelligence Services,” and other laws, administrative regulations, and departmental rules, in order to promote the healthy development of artificial intelligence, regulate the labeling of AI-generated or synthetic content, protect the lawful rights and interests of citizens, legal persons, and other organizations, and safeguard the public interest of society.

Article 2. These Measures apply where network information service providers that fall within the circumstances provided for in the “Provisions on the Administration of Algorithm Recommendation for Internet Information Services,” the “Provisions on the Administration of Deep Synthesis of Internet Information Services,” or the “Interim Measures for the Management of Generative Artificial Intelligence Services” (hereinafter referred to as “service providers”) carry out the labeling of AI-generated or synthetic content.

Article 3. “AI-generated or synthetic content” refers to information such as text, images, audio, video, and virtual scenes that is generated or synthesized using artificial intelligence technology.

Labels for AI-generated or synthetic content include explicit labels and implicit labels.

“Explicit labels” refers to labels added to generated or synthetic content, or to the interface of an interactive scenario, that are presented by means such as text, sound, or graphics and that users can clearly perceive.

“Implicit labels” refers to labels added to the file data of generated or synthetic content through technical measures, which users cannot readily perceive.

Article 4. Where the generation or synthesis services provided by a service provider fall within the circumstances of Article 17, paragraph 1 of the “Provisions on the Administration of Deep Synthesis of Internet Information Services,” the service provider shall add explicit labels to the generated or synthetic content in accordance with the following requirements:

(1) For text, add labels such as a text prompt or a general symbol prompt at the beginning, at the end, or at an appropriate position in the middle of the text, or add a conspicuous prompt label in the interface of the interactive scenario or around the text;

(2) For audio, add labels such as a voice prompt or an audio rhythm prompt at the beginning, at the end, or at an appropriate position in the middle of the audio, or add a conspicuous prompt label in the interface of the interactive scenario;

(3) For images, add a conspicuous prompt label at an appropriate position on the image;

(4) For video, add a conspicuous prompt label on the opening frame of the video and at an appropriate position around the video during playback; conspicuous prompt labels may also be added at the end and at an appropriate position in the middle of the video;

(5) When presenting a virtual scene, add a conspicuous prompt label at an appropriate position on the opening screen; conspicuous prompt labels may also be added at appropriate positions while the virtual-scene service continues;

(6) In other generation or synthesis service scenarios, add conspicuous prompt labels suited to the characteristics of the application concerned.

Where a service provider offers functions such as downloading, copying, or exporting generated or synthetic content, it shall ensure that the file contains explicit labels that meet these requirements.

Article 5. Service providers shall, in accordance with Article 16 of the “Provisions on the Administration of Deep Synthesis of Internet Information Services,” add implicit labels to the file metadata of generated or synthetic content. Implicit labels shall contain production-element information such as the attribute information of the generated or synthetic content, the name or code of the service provider, and the content number.

Service providers are encouraged to add implicit labels to generated or synthetic content in forms such as digital watermarks.

“File metadata” refers to descriptive information embedded in the file header according to a specific encoding format, used to record information such as the file’s source, attributes, and purpose.

Article 6. Service providers that provide network information content dissemination services shall take the following measures to regulate the dissemination of generated or synthetic content:

(1) Verify whether the file metadata contains an implicit label; where the file metadata clearly indicates that the content is generated or synthetic, add a conspicuous prompt label around the published content by appropriate means, clearly reminding the public that the content is generated or synthetic;

(2) Where no implicit label is verified in the file metadata but the user declares the content to be generated or synthetic, add a conspicuous prompt label around the published content by appropriate means, reminding the public that the content may be generated or synthetic;

(3) Where no implicit label is verified in the file metadata and the user has not declared the content to be generated or synthetic, but the service provider providing network information content dissemination services detects an explicit label or other traces of generation or synthesis, identify the content as suspected generated or synthetic content and add a conspicuous prompt label around the published content by appropriate means, reminding the public that the content is suspected to be generated or synthetic;

(4) Provide the necessary labeling functions, and remind users to proactively declare whether the content they publish contains generated or synthetic content.

In the circumstances in items (1) through (3) of the preceding paragraph, dissemination-element information such as the attribute information of the generated or synthetic content, the name or code of the dissemination platform, and the content number shall be added to the file metadata.

Article 7. When conducting review for the listing or launch of an application, an internet application distribution platform shall require the internet application service provider to state whether it provides AI generation or synthesis services. Where the internet application service provider provides AI generation or synthesis services, the internet application distribution platform shall verify the materials relating to its labeling of generated or synthetic content.

Article 8. Service providers shall clearly set out, in their user service agreements, the methods, styles, and other specifications for labeling generated or synthetic content, and shall prompt users to read and understand the relevant labeling management requirements carefully.

Article 9. Where a user asks a service provider to provide generated or synthetic content without explicit labels, the service provider may, after setting out the user’s labeling obligations and responsibilities for use in the user agreement, provide generated or synthetic content without explicit labels, and shall retain, in accordance with the law, relevant logs such as information on the recipient for no less than six months.

Article 10. Users who publish generated or synthetic content through network information content dissemination services shall proactively declare it and label it using the labeling functions provided by the service provider.

No organization or individual may maliciously delete, tamper with, forge, or conceal the labels for generated or synthetic content provided for in these Measures, provide tools or services for others to carry out such malicious acts, or harm the lawful rights and interests of others through improper labeling means.

Article 11. Service providers carrying out labeling activities shall also comply with the requirements of relevant laws, administrative regulations, departmental rules, and mandatory national standards.

Article 12. When completing procedures such as algorithm filing and security assessment, service providers shall provide materials relating to the labeling of generated or synthetic content in accordance with these Measures, and shall strengthen the sharing of labeling information so as to support and assist in preventing and combating related illegal and criminal activities.

Article 13. Violations of these Measures shall be handled by the relevant competent departments, such as those for cyberspace affairs, telecommunications, public security, and radio and television, in accordance with their duties and with the provisions of relevant laws, administrative regulations, and departmental rules.

Article 14. These Measures shall take effect on September 1, 2025.



Closing

The original document was published in Chinese by the Cyberspace Administration of China (together with the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration); we translated it into English, which is what you read above. This translation is provided for quick comprehension only and should be used at your own discretion and risk. Always confirm the current requirements with qualified legal counsel.

If you need further help from our team, contact us today, and our experts will help you keep your presence in China compliant from the ground up.


Ready to make your app work in China?

Get Started Questions? Talk to an expert.

Frequently Asked Questions

Which AI services need a visible label under these Measures?

Article 4 requires the explicit label where a service falls under Article 17, paragraph 1 of the Deep Synthesis Provisions: services that could confuse or mislead the public, such as intelligent dialogue or writing that simulates a person, synthetic or imitated voices, face generation or replacement, and immersive realistic scenes. The implicit metadata label under Article 5 applies to generated or synthetic content generally.

What does Article 7 mean for an app distributed in China?

App stores have to ask, when an app is listed or launched, whether it provides AI generation or synthesis services, and verify the labeling materials of apps that do. That is why App Store Connect now asks every China mainland app whether it offers AIGC services, and Apple’s version of the check asks for the CAC filing details of the model behind the feature. Our guide to Apple’s AIGC declaration covers what it accepts.

Can users get AI output without the visible label?

Yes, on request. Under Article 9 the provider may supply content without the explicit label once its user agreement sets out the user’s labeling obligations and responsibilities, and it must keep the related logs, including information on who received the content, for at least six months. Article 9 does not lift the implicit label, and Article 10 requires users who publish such content to declare it and label it.

Is there a technical standard behind the Measures?

Yes. Article 11 requires labeling to meet mandatory national standards, and the standard written for it is GB 45438-2025 on labeling methods for AI-generated and synthetic content. It took effect on September 1, 2025, the same day as the Measures.

Ready to try 21YunBox?

Get Started