Associated Incidents
CCTV News: Recently, the People's Court of Xuanwu District, Nanjing, tried an "AI face-changing" fraud case. The defendant, Fu, used more than 1.95 million pieces of citizen personal information and AI face-changing software illegally obtained to log in to the victim's payment account and swiped his bank card for consumption.
Man invades other people's financial accounts through "AI face swapping"
After investigation, Fu illegally obtained more than 1.95 million pieces of citizens' personal information through online shopping, and then used these citizens' personal information to log in to the financial platform payment accounts of 23 victims through AI face-changing software. Among them, he illegally changed the payment password and linked mobile phone number of five people, used the bank card linked to the payment account of one person, and purchased two mobile phones, totaling RMB 15,996.
** Zhang Haiyue, Assistant Prosecutor of the People's Procuratorate of Xuanwu District, Nanjing City, Jiangsu Province: **Through technology fusion, AI generates a dynamic video, shaking left, right, mouth, such some financial videos may often do some commands when users log in, break through the defense line of financial payment platforms.
During this period, Fu tried to use AI face-changing to break through the "face recognition" authentication system of several financial payment platforms, but only succeeded in a financial payment platform.
** Xu Jia, prosecutor of the People's Procuratorate of Xuanwu District, Nanjing City, Jiangsu Province: ** He kept trying and tried many platforms, but other platforms may have had some loopholes in this platform because of the stricter risk control system.
[Nanjing, Jiangsu] Man sentenced to 4 years and 6 months for multiple crimes
The prosecutor's office believed that Fu's behavior not only constituted a criminal offense, but also seriously damaged the public interests of society, and filed a criminal incidental civil public interest lawsuit. In October 2024, the court sentenced Fu to 4 years and 6 months in prison for infringement of citizens' personal information and credit card fraud. Fu pleaded guilty and pleaded guilty. In addition, in response to the technical risk issues existing in a financial payment platform exposed in this case, the procuratorate promptly issued a legal risk reminder letter. At present, the financial payment platform has completed the rectification.
**Xu Jia: **The reason why this case happened is because only by brushing the face, a simple face brushing entered its platform, you can change the password, if there is an extra layer of identity authentication here, he has no way to achieve the change. No more continued crimes through AI face-changing have been found.
** "AI face swapping" can bypass the face recognition defense line? **
With the continuous upgrading of AI face-changing, identity forging and other technologies, is the way of facial recognition authentication of identity safe enough?
**Permission from Director of the Center for Digital Economy and Legal Innovation Research at the University of Foreign Economics and Trade: **We insiders in cybersecurity will say that no network is always secure, and each update and iteration may bring new vulnerabilities.
Improving Reliability of Face Recognition through AI Governance AI
Although vulnerabilities are inevitable, there is no need to panic. The risks brought by the technology can be solved by the development of the technology itself. Through AI governance, the problem of facial recognition attacks we face can be solved.
**Zhu Meng, a practitioner in the field of AI security: **In fact, the detection rate or detection rate of the top technology in the industry is more than 95%, and he has been able to respond to such AI risks in practice. For example, when we look in one direction, both of your eyes will look to the left or to the right front, but AI generated, it may have both eyes looking from two angles.
Experts introduced that at present, some units applying face recognition technology have adopted similar anti-counterfeiting identification procedures to improve the reliability of face recognition systems.
**He Yanzhe, Deputy Director of the Evaluation Laboratory of the Cybersecurity Research Center of the China Institute of Electronic Technology Standardization: **For example, a certain latest smartphone, when you make a video call, it will have a function, such as whether to identify the face opposite is forged, in fact, some such technologies have been applied.
Units that use facial recognition should have stricter security measures
Experts believe that in addition to constantly improving technical means and enhancing anti-counterfeiting identification capabilities, personal information protection remains a very important link. On the one hand, ordinary people should raise their awareness of self-prevention and avoid disclosing personal information at will; on the other hand, units using facial recognition should also adopt more stringent security measures.