The "stars" in short videos may not be what you see. As AI face-changing and onomatopoeia technologies become increasingly mature, counterfeit accounts in the name of celebrities and their families have appeared frequently, making it difficult for netizens to distinguish between genuine and fake accounts. A few days ago, actor Wu Yue posted a post on social platforms in the early hours of the morning, saying that someone used AI face-changing technology to impersonate his own image and sell goods for profit, which attracted widespread attention.
The AI video mentioned in Wu Yue's post was posted by an account called "Sister Hua Beauty Diary". The video is marked with "Celebrity Style, Recommended by Wu Yue". The virtual image, which is suspected to be produced by AI, appears in the appearance and voice of "Wu Yue" and recommends a skin care product. The narration in the video also claims that "Wu Yue, in order to give back to his fans, has teamed up with a skin care product manufacturer to provide benefits to his fans." In addition to Wu Yue, the end of the video also shows Deng Yaping's "bringing goods" scene, with "Author's Note: Contains AI-generated content" marked below.
Actor Wu Yue: This happened the night before yesterday. My good friend sent me a video and asked me if I had done this ad. I was surprised at the time, and then quickly thought of something. A week ago, my good friend also asked me if I was advertising this product? I said no, so he sent a picture, which was an advertisement for this product. Really angry, very emotional. Because in my opinion, these audiences actually buy related products out of trust in you. When we buy something, we usually choose a brand, and there is a consideration, but maybe because a celebrity endorses it, they buy it because they trust this person; if they are deceived, I will feel very uncomfortable. So I posted an angry Weibo message at that time.
CCTV reporter Cao Yan: When communicating with colleagues, what is everyone’s common feeling?
Actor Wu Yue: We are very helpless, especially helpless. Because you said it is false, when you want to tell everyone about it, you need a lot of steps, and sometimes it is not completed and the complaint fails. But for those who make fakes, their time and energy are relatively simpler and easier, so they are really helpless. The account was blocked the next day. I feel really grateful. In this case, he will not continue to spread it.
CCTV reporter Cao Yan: After seeing the live broadcast of this account, did you choose to complain and report it?
Actor Wu Yue: I have thought about it. Because we have a team and are now communicating with lawyers. But I thought this thing was a bit slow. As a public figure, if you post something on Weibo, everyone will see it in a minute. I want to first tell people that I am not selling this thing. This is the most important original intention.
Netizens find it difficult to tell the truth from the false
Counterfeiters are stuck in the dilemma of repeated clarifications
Wu Yue's experience is not unique. As the threshold for AI face-changing and onomatopoeia technologies continues to decrease, the image and voice of public figures are more likely to be fraudulently used. Previously, actors such as Wen Zhengrong had also been "live broadcast" and "drugged" without their knowledge. This type of video capitalizes on the celebrity's influence and public trust, making it difficult for netizens to tell the truth from falsehoods, and also puts those being counterfeited in a dilemma of repeated clarifications and rights protection.
Actor Zhang Zhihua: If you use the actor's portrait, you are infringing on me. I think you can sue him and take up legal weapons. We are a society governed by the rule of law. I think we should enforce the rule of law against infringing activities such as AI videos as soon as possible, because I think it is trivial for an actor to resist alone.
Tong Ruixin, artistic director of Shanghai Film Actors Theater: I am very familiar with Wu Yue. The face and tone of voice are all Wu Yue, but Wu Yue is clarifying that it is not her. This is a very scary thing. We don’t want this kind of thing to ferment without a bottom line. I think we should improve the management system and not let some people continue doing it indefinitely just because it has economic benefits.
Fake celebrities bringing goods
How to "solve" the three major questions
Actor Wu Yue encountered AI counterfeiting and selling goods, and posted an article in the early morning saying it was "too bad"; previously, actor Wen Zhengrong saw an AI fake version of herself in the live broadcast room late at night. She entered the live broadcast room and asked: "You are Wen Zhengrong, then who am I?" She was blocked by the other party as soon as she finished speaking. So, why do AI face-changing and voice-cloning cases of fake celebrities selling goods happen frequently? Can a video be exempted from liability if it is marked "AI generated"? How to provide evidence to defend rights if you encounter infringement? In response to these three major questions, let’s take a look at what the experts have to say.
Video labeled "AI generated"
Can I freely use other people’s images and voices?
Experts point out that the infringing account uses AI to synthesize the actor's portrait and voice to carry out live broadcasts without the actor's permission, which is a typical infringement of personality rights. Even if the video is marked with the words AI generated, it is not a talisman of immunity. If you use another person's image or voice without authorization from the person concerned, even if the label is eye-catching enough, the nature of the infringement will not change.
Wang Qi, associate professor at Beihang University School of Law: Publishers of similar content have two parts of obligations. The first part is that they must follow the laws and regulations governing artificial intelligence and label the content generated and synthesized by artificial intelligence, which is what we often call deep synthesis, so that people know that it is generated by AI and is not real. He also has another obligation under civil law. If the video uses another person's personality identifier, including portrait, voice, etc., he must also obtain the consent of the portrait or voice right holder.
Why are these types of AI counterfeit videos banned repeatedly?
According to experts, in response to the chaos of AI content, regulatory authorities have issued the "Measures for Labeling of Synthetic Content Generated by Artificial Intelligence", "Regulations on the Management of Deep Synthesis of Internet Information Services", "Interim Measures for the Management of Generative Artificial Intelligence Services", etc., to provide detailed regulations on content labeling and service provider responsibilities. However, there are still practical difficulties in implementing the law. Nowadays, the threshold for getting started with AI tools has been significantly lowered, and the cost of producing counterfeit videos is extremely low. Some merchants make profits by forging celebrities' images to sell goods. Even if they are investigated and punished, the amount of compensation is often not high. The illegal profits far exceed the illegal costs, and infringements will occur repeatedly.
To curb this kind of chaos, we must not only consolidate the main management responsibilities of online platforms, but also the regulatory authorities need to increase their investigation and punishment efforts and severely punish illegal merchants in accordance with the law.
Wang Qi, associate professor at Beihang University School of Law: Verification of operators within the platform must be done beforehand, early warnings and inspections must be strengthened during the process, and the ability to proactively discover violations of laws and regulations must be increased, and complaints must be accepted in a timely manner afterwards. In addition, legal responsibilities must be increased, and civil responsibilities can be strengthened. For example, consumers can claim punitive damages in accordance with the law, and market supervision and management departments must implement responsibilities such as investigation and fines for illegal activities in accordance with the law, increasing the violation costs of violators.
Encountered AI counterfeiting
How to safeguard rights
Once it is discovered that the portrait or voice has been counterfeited and misappropriated by AI, the person concerned can immediately record and take screenshots, and completely fix all the evidence such as the infringing video and account homepage. You can either complain to the platform and request that the infringing content be removed from the shelves; you can also file a lawsuit and assert your rights through judicial channels. Many infringing videos are deliberately borderline and combine images that are similar but not similar, creating obstacles for rights holders to provide evidence. Experts remind that when defending rights, you don’t need to be obsessed with images that must be “100% similar”.
Wang Qi, associate professor at Beihang University School of Law: During the process of rights protection, the infringed party must prove that the relevant video can identify the person. According to current judicial policies and judicial standards, it is not required to be 100% similar. As long as the infringed person proves that the person can be identified according to the understanding of the relevant public in society or relevant groups in the industry, it has constituted an infringement of the person's personality rights and can claim legal protection.
