On March 14, 2025, the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration jointly issued the Measures for the Labeling of Artificial Intelligence-Generated and Synthetic Content (shortened as the Labeling Measures), effective from September 1, 2025. This is China's first mandatory regulation specifically targeting AIGC content—all video content generated using AI technology must undergo labeling management.
I. Who Needs to Comply with These Regulations
According to the regulatory framework of the Labeling Measures, the obligated parties fall into two categories:
Category 1: Deep Synthesis Service Providers (platforms and tool developers)
These are platforms and technology vendors that provide AI video generation services. This includes mainstream tools such as Baidu ERNIE Bot, Tencent Smart Creation, and Jianying Professional Edition—these service providers must display prominent notices stating “This content may be generated by artificial intelligence” at visible locations in their user interfaces, and submit labeling-related materials during algorithm registration.
Category 2: Deep Synthesis Service Users (content creators and businesses)
These are individuals and organizations that use AI technology to produce videos. This includes social media bloggers, MCN agencies, and brand enterprises—these parties need to add labels or embed text overlays on the publishing platform stating “This video was produced with the assistance of artificial intelligence.”
II. Legal Consequences of Non-Compliance
According to Article 13 of the Labeling Measures, violations shall be handled by competent authorities including cyberspace, industry and information technology, telecommunications, public security, and radio and television administrations in accordance with relevant laws and regulations. In other words, regulatory agencies can impose penalties under existing laws such as the Cybersecurity Law and the Administrative Measures for Internet Information Services—specific fine amounts depend on the severity of the violation.
Category 1: Warnings and Rectification Orders (first-time violations)
If a creator publishes AI-generated content without the required labeling, regulatory agencies will issue a written warning and require supplementary labeling. The handling at this stage is relatively lenient—the primary purpose is to remind rather than punish.
Category 2: Legal Penalties (serious violations or refusal to rectify)
If a creator fails to correct their behavior after receiving a warning, or uses AI-generated content to create false information for profit (such as forging celebrity endorsement videos to promote products), they will face more severe administrative penalties. For business entities, this may affect the annual review of their business licenses and the normal operation of their platform accounts.
III. Comparison of Three Mainstream Labeling Methods
Category 1: On-Screen Text Labeling (simplest and most direct)
Add the text “This video was produced with the assistance of artificial intelligence” at the end of the video or in a prominent location. The advantage is that viewers can see the notice directly; the disadvantage is that it may affect visual aesthetics—especially for commercial promotional videos pursuing a cinematic quality, this labeling method may appear unprofessional.
Category 2: Implicit Labeling (most technically standardized)
The Labeling Measures explicitly defines two main forms: explicit labeling and implicit labeling. Implicit labeling embeds AI generation labeling information into the video file's metadata—viewers need a player that supports this standard to view these tags. The advantage of this method is that the labeling information does not interfere with the visual presentation, but the disadvantage is that ordinary users may not be able to perceive these hidden labels.
Category 3: Platform-Side Label Display (most flexible and convenient)
Mainstream short video platforms such as Douyin, Xiaohongshu, and Bilibili have built-in AI content recognition features—when the system detects that a video contains AI-generated segments, it automatically adds an “AI-Generated” label on the playback interface. Creators only need to check the relevant option when uploading to complete the labeling, without modifying the original file.
IV. Recommended Compliance Workflow
Step 1: Clarify your role (individual creator / corporate brand / MCN agency); Step 2: Choose the labeling method according to platform requirements (on-screen text / implicit label embedding / platform-side label display); Step 3: Conduct a self-check before publishing—ensure all AI-generated segments are correctly labeled; Step 4: Retain original materials and production records for regulatory review.
V. Common Misconceptions Clarified
Misconception 1: “I don't need labeling if I use open-source tools”
Incorrect. The regulatory scope of the Labeling Measures covers “all entities that use AI technology to generate content”—whether you use commercial software or open-source models, as long as you produce video works intended for public dissemination, you must apply labeling.
Misconception 2: “Posting to my Moments doesn't count as public dissemination”
Correct. If a video is only shared within private social circles (such as WeChat Moments or family group chats), it is generally not considered public dissemination—however, once published to open platforms such as Douyin or Xiaohongshu, it triggers compliance obligations.
VI. Future Trend Predictions
According to predictions from industry analysis agencies, more detailed implementation standards will be issued in the second half of 2026—potentially including a classification and grading management system for AI-generated content and a cross-platform joint regulatory mechanism. Creators are advised to stay informed about policy developments and adjust their workflows accordingly.
