AI-generated videos can now evade facial recognition systems; Hangzhou Intermediate People's Court case and multiple AI fraud cases exposed

On October 26, the topic #AI-generated videos can now evade facial recognition systems# trended on social media, drawing widespread attention. According to a CCTV.com report, the Hangzhou Intermediate People's Court published a case this year in which AI-generated videos were able to evade facial recognition systems. Two criminals collected photos of others online and used AI to generate dynamic videos of nodding, shaking the head, blinking, and so on; then, by tampering with camera data, they successfully bypassed the facial verification of a shopping platform and stole personal information from accounts. Abuse of AI technology to pass off fakes as real and commit fraud. Currently, AI-generated videos are making it increasingly hard for people to tell real from fake, and many people use technologies such as AI face-swapping and voice cloning to generate fake videos for fraud. On October 18, actor Wu Yue spoke out publicly, angrily denouncing an account for using AI to impersonate her image and voice to promote skincare products. According to screenshots retained by Wu Yue, the product had cumulative sales of 16,492 units, and a rough estimate put the sales amount at more than 1.3 million yuan. At the end of the counterfeit video, an image imitating Olympic champion Deng Yaping even appeared to continue promoting products. According to a CCTV Finance report, the Yuhua District People's Court in Changsha recently heard a case involving the use of AI-generated fake face videos to fraudulently use bank cards. A university student bought "AI face forgery video" technology on an illegal website, produced dynamic videos of other people's faces, reset payment passwords, and in less than four months fraudulently used bank cards to take more than 50,000 yuan. Why does AI fraud keep recurring despite repeated bans? Experts analyze: "Dozens of yuan and a dozen seconds of audio are enough to 'replicate' anyone; the barrier to entry is extremely low, templates can be copied in bulk, and when an account is banned they just switch shells—this is the root cause of the persistent chaos." How can we guard against AI fraud? New types of AI fraud are emerging endlessly and becoming increasingly deceptive. In daily life, how should we protect ourselves? 1. During video calls, ask the other person to make specified movements (for example, blink three times, touch their nose, or wave a finger or other object in front of their face). If the image shows unnatural tiny changes such as delays or abnormal distortion, the other person is very likely using "AI face-swapping" technology. 2. Unless necessary, do not provide important information such as ID card numbers, workplace, home address, or job title to strangers, and avoid leaving faces, photos, voices, fingerprints, and other such data on websites and mini-programs as much as possible. 3. When receiving a call from an unknown number, if the other person keeps silent, do not speak first either; wait quietly for three seconds. If the other person still does not make a sound, hang up promptly. If after connecting it is an AI robot impersonating a so-called customer service agent, likewise do not speak and hang up immediately. Compiled from China News Service, CCTV News, CCTV.com, CCTV Finance, Global Times, People's Daily, Jing Shi Live, and others.