Just one.photo facial reconstructionIn the last few days, the "N" version of himself has been created on the Internet, and the "N" version of himself has been created on the Internet, and the "N" version of himself has been created on the Internet.Deepfake"Topics related to the (deep forgery techniques) incident surged to the top of the search engines, utilizing "AI face-changing"Infringement scenarios abound, posing greater cybersecurity risks.
"AI face change" set of tricks, ordinary users should be how to identify the authenticity? On September 8, the opening of the 2024 National Cybersecurity Awareness Week, the Southern Finance and Economy all-media reporter experienced a "second face" of "black science and technology", and "AI face" on how to comply with the use of technology, how to prevent the risk of "AI face" and other issues. How to comply with the use of technology, how to prevent the risk of "AI face change" and other issues, interviewed Zhang Bo, a member of the Artificial Intelligence Security Governance Specialized Committee of the China Cyberspace Security Association, and assistant president of Tianrongxin Technology Group.
Southern Finance: What is the principle of the "AI face-swap" technology?
Zhang Bo:The process of "AI face change" is divided into three steps, the first is the face recognition and tracking, using deep learning algorithms to accurately recognize the face image in the video, followed by the extraction of facial features, the collection of the eyes, nose, mouth of the face in the image of these key facial features, and match these features with the target face image, and the third is the fusion of face changes. The third is the fusion of changes in the face.
At present, the "AI face" technology is more mature, as long as we get a character face photo, you can realize the face, but also can do real-time video calls to change the face, to the extent that the fake is real, some of the naked eye alone is difficult to distinguish. In addition to changing faces, if combined with a change in the background environment, change the audio, the degree of realism is even higher.
SCF: What are the possible dangers of misuse of facial information?
Zhang Bo:The misuse of face information, for the individual, light is damage to reputation, the image of the spoof, if the identity is counterfeited, the serious case of property will be damaged. I think the application of a technology is a "double-edged sword", both positive and negative, and need to be guided by the implementation of laws, regulations and systems to standardize, but also need to enhance security awareness.
Southern Finance: How to prevent the risk of "AI for Face"?
Zhang Bo:There are several aspects. The first is to raise one's awareness of security risks. "AI face-swapping" may appear in conjunction with transaction fraud scenarios, so when it comes to sensitive scenarios such as providing personal information and transferring money, one should be highly vigilant in order to prevent being cheated. Second, when receiving video information such as video calls, carefully observe whether the light background, facial contours and other details in the other party's video are natural, and if necessary, ask the other party to make rapid movements such as raising their head, lowering their head, and turning their head from side to side, to observe whether there are any anomalies in the image, and further verify the authenticity of the video. Third, when faced with financial transactions such as money transfers and remittances, you can confirm the identity of the other party through other relatively reliable means, such as making a phone call in person, to ensure that the transaction information is correct and to avoid losses due to gullible belief in false information.
SCF: What should you do if you find out you've been face-swapped?
Zhang Bo:First of all try to do a good job of prevention, because now the process of changing face through a photo is very simple, there are many tools are open source. Strengthen the precautionary protection of personal information, and be vigilant about the need for sensitive information such as face information and fingerprint information, so as not to easily leak it. At the technical level, artificial intelligence detection can be used to identify whether the video content has been forged, but also increase the authentication and digital watermarking to increase the difficulty of forgery. In addition, from the perspective of security protection, it is necessary to strengthen the implementation and enforcement of laws and regulations and the corresponding regulatory system to regulate the use of "AI face-swapping" technology and application scenarios. In terms of law, the need to accelerate the development of relevant laws and regulations and regulatory measures and landing, if the photo is found to be imitation, the need to have some legal avenues to complain, while also need to further strengthen the public's security awareness and prevention of fraud publicity, to improve the understanding of false fraudulent information and alertness.
Southern Finance: In what areas can "AI for Face" technology be used for compliance?
Zhang Bo:Nowadays, it is more widely used in the entertainment industry, and there are also some commercial avenues, for example, it can allow you to produce a photo album without leaving your home, which is actually a good thing. Although this technology is called face-swapping, but in fact, through this technology can also be changed into something else, for everyone's entertainment. Only it is really a "double-edged sword", if really by the ulterior motives of the people, the use of lawless elements may cause great harm, so the regulation of compliance is indeed very important. Users using "AI face-swapping" technology should try to choose products launched by large companies and platforms, and try not to upload their personal information through unknown channels.