Face changing software unblocks tens of thousands of account fraud Can face recognition be used as a decisive verification method?

On October 18, according to the official news of the Jiangsu Internet Police, the Jiangsu police cracked a case of "gangs using face-changing Apps to unseal tens of thousands of restricted Micro-signals". The criminal gang used a number of mobile phone software to forge faces, passed the WeChat identification and authentication cancellation mechanism, and "resurrected" a number of WeChat accounts that were restricted from logging in due to violations, helping fraud gangs to use these WeChat accounts to carry out fraud.

The case has raised questions about the security of face recognition technology by some netizens. He Yanzhe, deputy director of the Evaluation Laboratory of the Network Security Research Center of the China Electronics Technology Standardization Institute, told Nandu reporters that the use of technology to crack face recognition cases will occur, in fact, not because face recognition technology can not be safe enough, but to balance accuracy and ease of use, so it may appropriately reduce the elements of security verification. He also reminded that face recognition should be an auxiliary means, and if it is used as a decisive verification method, it is likely to go wrong.

Criminal gangs used the face-changing App to unblock tens of thousands of restricted Micro-signals

According to @Jiangsu Internet Police Weibo, when the police investigated the case, they found that the people involved in the case used a WeChat signal that had been disabled due to fraudulent activities. After investigation by the police, the people who registered these WeChat accounts were from remote areas, and they sold the accounts to others after registering The WeChat accounts. And what buyers do with these WeChat signals, the registrant is not clear.

The police found that the restarted WeChat was used by overseas fraudsters, and the key to lifting the restriction on login was face verification. In the end, the police dug up a black industrial chain that re-enabled restricted accounts, and successfully destroyed this illegal and criminal gang suspected of illegally obtaining citizens' personal information and forging face recognition authentication.

According to the police handling the case, the suspect used nine software, and after putting the picture into the software for a series of operations, a dynamic video picture will be formed. The suspect uses such video pictures to conduct face verification, lift the restriction on login, and help the fraud gang commit crimes.

According to the confession of the criminal suspect, the fraud gang lifted the restrictions on a WeChat account and was able to obtain 100 yuan, and after review, the fraud gang restarted more than 10,000 WeChat accounts and illegally profited more than 1 million yuan.

In an interview with the Jiangsu Public News Channel, Wang Qiangwei, deputy captain of the network security protection brigade of the Nantong Rugao City Public Security Bureau who cracked the case, said: "Software vendors are required to continuously optimize the algorithm and improve the security capabilities of the software itself. ”

Is face recognition technology safe? Expert: It should not be used as a decisive verification method

Judging from the details of the case-handling personnel revealing that "the software is used to generate pictures and videos", the technology used by the suspects is not unusual, but a deep-depth forgery technology that has been relatively mature. Deep forgery is an artificial intelligence technology that uses neural network technology to learn large samples and synthesize existing images and videos onto target images or videos. At present, deep forgery technology is often used for AI face swapping, in addition to voice simulation, video generation and so on.

Deep fake technology of faces first appeared in 2016, this technology began with the magical fresh AI face swap effect, but by 2017, pornographic videos synthesized by deep fake technology had appeared on the Internet, and since then, deep fake works have been banned by some foreign social networking sites and video sites. But fake celebrity pornographic videos have gradually developed into a black industry chain, and celebrities have been victimized repeatedly.

Face recognition is another hardest hit area for deep forgery crime. Nandu reporters have investigated the "AI face-changing" chaos on QQ, and found that many people who provide AI face-changing services are also doing "cracking face recognition" business.

In this case, wechat face recognition can be cracked with photos, making many netizens question whether face recognition technology is really reliable.

He Yanzhe told Nandu reporters that the security of face recognition is often discussed and questioned by everyone, but in fact, it is not that the face recognition technology cannot be safe enough, but it is necessary to balance accuracy and ease of use, so it may appropriately reduce the elements of security verification.

"If face recognition was supposed to test ten points, but now discounted, only five points, naturally there is no test ten points of security, but the convenience will rise." He said, "Otherwise, maybe you wear a hat, or the light changes, and you won't be able to recognize it." ”

A head face recognition technology provider also told Nandu reporter that under different application scenarios, the security level of face recognition is different, and the accuracy will indeed be different. For example, mobile phone unlocking more pursuit of speed and convenience, the accuracy of recognition may be affected to a certain extent; and in the large payment scenario of financial institutions, on the one hand, there will be live detection and other technologies to strengthen security, on the other hand, usually through the device code, mobile phone tail number, password and other information cross-verification.

He Yanzhe believes that face recognition should have been an auxiliary means, and if it is used as a decisive verification method, it is likely to go wrong.

In addition, in the face of netizens' doubts about the security of face recognition, he proposed that when discussing the security of face recognition, a question should be considered - face recognition, and no face recognition, which is more secure? In his view, in the era of passwords as a verification method, the phenomenon of social account theft also exists. And face recognition, most of the time, still creates a certain threshold. "I can't say that my account was stolen because of face recognition. This is not a complete logic. He said.

Written by: Nandu reporter Li Yaning