AI-Generated Content Misuse and Labeling Gaps in China
Unlabeled and misused AI-generated content in China, including fake celebrity endorsement videos, has deceived consumers and elderly persons into financial losses. Regulators have now issued mandatory labeling requirements for all AI-generated content, effective September 2025, though reports indicate enforcement gaps persist on some platforms.
What happened
A company in Beijing's Haidian District used AI technology to edit videos of a well-known television presenter, inserting false spoken endorsements and publishing the manipulated videos on its own social media accounts as short video advertisements. The ads promoted an ordinary food product (deep-sea polyunsaturated fish oil) with false medical claims of treating dizziness and headaches, violating advertising regulations.
AI-generated content without clear labeling has been used to create fake endorsement videos that deceive elderly consumers. Some elderly persons, unable to verify the authenticity, made purchases based on these fabricated recommendations and fell victim to scams.
The Chinese government issued the 'Artificial Intelligence Generated Synthetic Content Marking Method' which took effect on September 1, 2025, requiring all AI-generated text, images, audio, and video content to be clearly labeled with both explicit and implicit identifiers.
Reported impact
Elderly consumers were misled into purchasing products recommended in fake AI-generated videos, resulting in financial losses and falling victim to fraudulent schemes. The widespread nature and scale of such incidents remain unclear from available reports.
The proliferation of unlabeled AI-generated content has created challenges for consumers attempting to distinguish authentic from synthetic information, undermining trust in online content and complicating media literacy efforts.
Index analysis
The failure of platforms to mandate and enforce labeling of AI-generated content enabled the continued circulation of deceptive synthetic media. Although regulatory frameworks were issued, reports indicate that as of October 2025, some short video platforms and livestream services still hosted unlabeled AI content.
The misuse reflects gaps in content moderation and the absence of adequate mandatory disclosure mechanisms prior to regulatory action, allowing bad actors to exploit synthetic media technology at scale without sufficient friction or detection.
Governance controls
China's four-part regulatory framework issued on September 1, 2025 mandates explicit labels (text, sound, or graphics clearly visible to users) and implicit labels (metadata embedded in content files) for all AI-generated content. Platforms including Douyin, Kuaishou, Xiaohongshu, Bilibili, and Ximalaya have announced implementation of labeling and detection systems.
The regulations assign responsibility across the content supply chain, requiring generative AI service providers to add labels, content distribution platforms to verify metadata, and app distribution platforms to audit compliance during application review. Regulatory agencies have authority to impose administrative sanctions for non-compliance.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.