AI-generated deepfake livestreams used to fraudulently sell counterfeit goods in China
Merchants in China have used AI deepfake video systems to create fraudulent e-commerce livestreams that impersonate celebrities, academics, and Olympic athletes, and falsely promote counterfeit branded products including protected geographical indication peppers. Consumers unknowingly purchase based on false endorsements and misrepresented product origins.
What happened
Merchants on Chinese short-video and livestreaming platforms have used AI technology to create deepfake videos for e-commerce fraud. Some deepfakes impersonate celebrity endorsers including Olympic champions, film and television stars, and academic experts; others fraudulently promote counterfeit branded products such as Zhuangshugang peppers, a protected geographical indication product from Hunan.
In livestreams selling Zhuangshugang pepper seeds, AI-generated background videos showed physically impossible scenarios such as farmers harvesting peppers while standing in deep water, or objects merging together in ways that violate physical laws—flaws known as 'clipping' or 穿模. An AI video generation system can produce a convincing deepfake livestream video from a single photograph in minutes, changing clothing and props in the process.
Reported impact
Consumers and fans struggle to distinguish authentic from fake livestream content, purchase products based on false assumptions about their origin or endorser, and report being defrauded. Professor Li Ling from Peking University's National School of Development publicly stated that deepfake videos falsely impersonating her had been used to fraudulently promote medical books and treatment methods she had never endorsed.
The fraud damages the market reputation of authentic Zhuangshugang peppers and violates the personality rights of impersonated public figures by using their likenesses and voices without consent.
Index analysis
China's National Cyberspace Administration and three other government departments implemented the 'Artificial Intelligence Generated Content Labeling Measures' on September 1, 2025, requiring all AI-generated text, images, and videos to carry explicit and implicit labels. However, some merchants circumvent platform detection systems by tampering with data and converting file encoding before uploading content.
Legal experts stated the conduct involves suspected civil infringement and potential criminal violations of consumer protection law, e-commerce law, and seed law, as merchants misrepresent product origins and properties to deceive consumers.
Governance controls
Major online platforms have upgraded AI content detection systems to require labeling of AI-generated works, but detection remains imperfect. Legal experts recommended platforms use digital watermarking and blockchain-based evidence preservation to strengthen authentication and traceability, while calling for refined legal standards for labeling, audit, and accountability mechanisms.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.