Tencent's live-commerce moderation now faces a two-month CAC campaign, launched September 30, that bars platforms from treating AI labels as cover for celebrity impersonation in sales videos.
Tencent’s WeChat Channels has permanently banned a skincare-selling account that used an AI-generated likeness and voice of actor Wu Yue (吴越) to pitch its product, after she publicly condemned it in the early hours of October 18. The Paper reported that the WeChat team told its reporter the account, called 花姐美肤日记 ("Sister Hua’s skin diary"), had been banned for good and that its content had been cleared and could no longer be viewed. The same account’s videos also featured a suspected synthetic version of table tennis world champion Deng Yaping.
The case exposes a gap in China’s labeling approach. The video promoting the product carried the platform notice "author’s note: contains AI-generated content," yet it still put words in a real person’s mouth. China has required such labels since September 2025, and the country’s internet regulator opened a nationwide campaign against AI impersonation of public figures less than three weeks before Wu Yue’s post. How much the account sold, how long the videos ran and who made them were not disclosed in any of the reports reviewed.
For investors following Tencent, the episode lands on live commerce inside WeChat. TechNode reported in early 2023 that WeChat Channels, Tencent’s short-video product, had grown gross merchandise value from livestream shopping by 800% in 2022, according to figures released at its annual event. A takedown that came only after a celebrity complained publicly is the kind of case regulators now say they want platforms to catch earlier.
The account sold skincare with a cloned voice and a pregnancy claim
Wu Yue posted on Weibo shortly after waking, asking who was using AI to "grab people" and do harm in their name, and demanding that "Sister Hua" return the money taken from buyers. Beijing Daily put the time of her post at about 3 a.m. and reported that the account could no longer be found through WeChat search.
According to The Paper, the video opened with a face-swapped livestream sales segment, then cut to real footage of Wu Yue at public events, laid over with a voice imitating hers. The fake voice claimed she had approached the brand herself, and said the results were plain to see. The product was a serum, according to Beijing Daily, and Guancha reported that the account marketed it as something a celebrity uses and Wu Yue recommends, with the synthetic voice adding that even pregnant mothers could use it. 21jingji reported that the pitch framed the product as a reward for longtime fans, arranged through a supposed tie-up with the manufacturer.
The Paper noted that WeChat Channels announced in March that it had set up routine controls for AI-generated content and was adding labels to suspected synthetic posts so users could identify them. In this case the label was in place, and the account still traded on a real person’s identity.
Regulators have already said a label does not excuse impersonation
The Measures for Labeling AI-Generated Synthetic Content (人工智能生成合成内容标识办法), published by the Cyberspace Administration of China (CAC) with three other agencies in March 2025 and in force since September 1, 2025, require services to add labels that users can clearly perceive, in text, sound or graphics, and require platforms that distribute content to flag material identified as AI-generated. They are a disclosure regime: they say content must be marked, and they bar anyone from deleting or forging the marks.
The CAC’s newer campaign notice, issued on September 30, 2026, goes further. The two-month drive, titled 清朗·整治短视频虚假人设乱象 and aimed at fake personas on short-video platforms, targets accounts that use AI-synthesized audio and video to impersonate public figures for fake endorsements and marketing. The CAC campaign tells platforms they may not rely on tags such as "AI-generated" or "fictional performance" to sidestep content governance. Wu Yue’s case fits that description closely.
Courts are moving in the same direction. The Supreme People’s Court (SPC) issued an opinion on September 7 stating that courts will support claims from people whose names or likenesses are used without consent to generate identifiable digital figures, and gives similar protection to a person’s voice against unauthorized AI imitation. On October 16, the Beijing Internet Court described a case in which a livestream host used AI to synthesize an elderly person’s voice for promotion; The Paper’s court report said the merchant that commissioned the promotion was held jointly liable for failing to review it.
Platforms face pressure to find fakes before the victims do
Rival platforms report the same problem at scale. Guancha reported that Douyin’s e-commerce arm had handled about 11,000 impersonating creator accounts and more than 160,000 pieces of infringing content by March this year. Beijing Daily listed other actors who have complained of AI-made sales videos using their faces, and argued that platforms should screen such content before consumers are deceived, rather than act only after a celebrity speaks out.
Neither Wu Yue nor Deng Yaping has announced legal action in the reports reviewed, but the SPC opinion and the Beijing court’s reasoning on merchant liability set out how such claims are treated, including against merchants who commission promotions. The CAC campaign has roughly six weeks left to run from its September 30 start. The development to watch is whether the CAC names WeChat Channels in its campaign results, or whether Tencent publishes enforcement figures for AI impersonation on Channels of the kind Douyin has already released.