In recent years in China, there has been a proliferation of videos using AI technology to deeply synthesize well-known figures, swapping faces and voices. Recently, actress Wen Zhengrong revealed that after finishing filming around 3 or 4 in the morning, she stumbled upon a live stream where someone was impersonating her using AI to promote products, and when she approached to inquire, she was blocked by the individual.
According to a report by CCTV, the state media of the Chinese Communist Party, multiple live streams featuring “Wen Zhengrong” are not actually her, as these contents are all fabricated through AI to create false live streams, piracy, and rebroadcasting of past clips.
Wen Zhengrong expressed, “In fact, these products were never recommended by us, nor have we ever sold them.” She mentioned that as early as before April, she had been mentioned in various live streams during the same time period with different makeup and hairstyles promoting different products but saying the same things.
Wen Zhengrong pointed out that fans would trust her and make purchases, but these live streams were using AI to confuse visuals and audio, making it difficult for her to prove her authenticity.
In response, the platform announced on the evening of November 5th that they have taken action by removing over 10,000 infringing videos impersonating Wen Zhengrong, with more than 6,900 of them involved in promoting products, and penalized 37 accounts for violations.
Wen Zhengrong being impersonated by AI for product promotions is not an isolated case. In August, according to CCTV News, on a social media platform, a blogger used AI to mimic the voice of Olympic champion Quan Hongchan to sell free-range eggs. Similarly, the voices of Sun Yingsha and Wang Chuqin were also “cloned” to promote “Quanbao Family Free-range Eggs”, leading many netizens to mistakenly believe it was the real Quan Hongchan and proceed to place orders.
Furthermore, there are cases of bloggers “cloning” the voices of actors like Jin Dong to conduct 24-hour live broadcasts through voice imitation technology.
In recent years, the Chinese Communist Party has been vigorously developing artificial intelligence (AI). With the continuous advancement of the AI industry, AI technology and tools have become a means for some businesses to “ride the wave” of traffic. Public figures such as actors, athletes, hosts, scholars, and professors, due to their high visibility and the trust of their fans, have become prime targets for AI impersonation. The relatively low-cost AI marketing model has been frequently replicated and difficult to put a stop to by these unlawful businesses.
Public information indicates that a large number of AI tools and tutorials are easily accessible on online platforms. The abuse of AI technology to impersonate celebrities for product sales has given rise to a black market chain of “infringement, fraud, and profiteering”.
In fact, Wen Zhengrong’s team had already compiled relevant evidence and sent legal letters to these unlawful businesses long before. Fans also disclosed that they had reported 18 “Wen Zhengrong” live streams in a single day, but the problem persisted as new streams kept popping up despite their efforts to report them.
Some of these pirate accounts use techniques like “mirror flipping” and “picture-in-picture blocking” to evade platform AI recognition systems, easily circumventing bans by changing IP addresses, using virtual locations, and modifying device IDs.
According to analysis by legal experts, these piracy activities often spread across different platforms, with user behavior data not shared between platforms, making it challenging to track the complete chain of infringements.