News

Face recognition causes many privacy violations

Recently, Clearview AI, an American face recognition company, has been exposed to data. The American startup company has captured about 3 billion face photos from social network platforms such as Facebook, twitter, instagram, LinkedIn, venmo and Youtube without the permission of the parties. The main customer is the police station, which allegedly can identify the suspect’s face information in a few seconds. Companies like Twitter, Google, YouTube have sent letters to Clearview AI asking them to stop accessing their own website data. The rather low-key company came to the public for alleged privacy violations.

Why can’t privacy be violated?

Some people will say that the starting point of using face recognition and big data monitoring is good, and the motivation is good. The CEO of Clearview AI, Juan ton, argued that Clearview AI is designed to help law enforcement agencies better combat criminal crime. He said he had “the best intentions.”. The problem is that, under the good intentions, the right to process privacy data is concentrated in the hands of a small number of people. It is very dangerous. Besides the risk of leakage, there are risks of abuse. For example, the police officers who frequently disclose use their posts to find the hotel occupancy records of a person. It is the weakness of human nature to pry into privacy and immoral curiosity. People will also have desires to seek their own interests. We can not prevent them.

How do we face privacy violations?

In fact, technology is neutral, and the person behind it is responsible for evil. So we should limit the possibility of doing evil by law and other means. For example, the San Francisco parliament passed the stop secret Surveillance Ordinance, which explicitly prohibits the use of face recognition technology by San Francisco law enforcement. It is the first time in history that the government has been explicitly “prohibited” from using such surveillance technology as face recognition, because the government departments are most likely to abuse public power.

We cannot deny the power of technology, what we should do is prevent such power from being abused. In the collection of privacy information such as face recognition, some principles should be met, such as only collecting the least information used, the user has the right to know before collecting and obtaining consent. The risk of information storage and processing should be controlled, and remedial measures should be taken. In case of any problem, who will go for the bottom, the responsibility shall be clear. It is a good face recognition to be able to achieve these face recognition.

Why is face recognition more sensitive in the field of personal information protection?

The sensitivity of face recognition technology comes from the fact that we can hardly modify our facial information. Other types of personal information, such as user name, email, mobile phone number, and even name, can be easily changed. Once the facial recognition information collected by face recognition technology is collected, it may be permanently collected.

With the awakening of public privacy protection awareness, the application of face recognition technology is making more and more people alert. On the one hand, this vigilance comes from the unpleasant caused by peeping, and people in many scenes don’t know how their facial feature information will be used; On the other hand, people have no choice but to offer facial features in many scenes, and they have to offer their “face” unconditionally.

What is a good way to get “consent” in public places?

Public place is the most difficult scene for face recognition technology to obtain “consent”, which can be divided into two parts: notification and consent.

When collecting personal information in public places such as automobile station, railway station, subway station, airport and shopping mall, it is suggested that the personal information controller display and inform the personal information subject in a significant way. For example, the store posted a notice: “the mall has a face recognition system to analyze passenger flow or make personalized recommendations. We promise to protect your face and other information security. Please consult the inquiry desk or scan QR code for details. ” “Consent” is achieved by providing a choice scheme not to collect personal information, for example, setting up a purchase method that does not become a member; Payment method without face recognition.

In addition, in the guidance on personal data processing through video equipment issued by the European Commission for data protection (3 / 2019), it is recommended to use double-layer notification structure. The first layer Posts notices in public places where video processing personal data is used, providing brief description, and provides access to the second layer description through two-dimensional code and other channels; The second layer is explained in detail.

Of course, getting “consent” requires design in specific business scenarios based on actual conditions, and balancing legal compliance with business needs.

What measures can be taken in addition to obtaining consent?

In principle, the original personal biometric information (such as portrait photo) should not be stored, but only the summary information processed by the algorithm should be stored. That is, according to the principle of “minimizing” personal information protection, the life cycle of facial recognition features is shortened, and the legal risk of face recognition technology can be reduced by reducing the collection of data.

“Face recognition at airports: what are the challenges and the main principles to follow?” published by cnil, a French data protection agency In the report, the technical measures required: (1) only after the relevant passengers make the action, can the face recognition camera be activated（ 2) Through technical configuration, blur the face of other passengers in the background（ 3) The billboards and ground marks of control area and traditional control area are distinguished by facial recognition.

With the development of face recognition technology, what problems should we pay attention to forward-looking?

In September 2020, Professor Li Feifei, a professor of computer science at Stanford University, and Professor Arnold mills stein of Stanford Medical College, published a paper entitled “lighting up the dark space of medical treatment with environmental intelligence” in nature, which emphasized the potential of environmental intelligence technology. Environmental intelligence, which is the electronic space that can respond and feedback to human existence through machine learning and contactless sensors. Face recognition technology is an important part of environmental intelligence.

In addition to the use of cameras, environmental intelligence technology also uses depth sensor, temperature sensor, wireless sensor, sound sensor and other equipment. Environmental intelligence can be used in intensive care, clinical care, and even the elderly to live independently. Because the essence of environmental intelligence is comprehensive monitoring, the depth and breadth of data collection is far beyond the face recognition technology, so the environmental intelligence technology must pay more attention to privacy and data security design.

In recent years, face recognition technology, environment intelligence technology, automatic driving and other technologies have developed rapidly, behind which is the collection and processing of massive data. The processing of these data has a profound impact on the relationship between everyone and technology. While enjoying the convenience brought by the technological progress, we need to pay attention to the protection of personal information, privacy and portrait by new technology; While enjoying the profit brought by technological progress, operators should pay attention to how to reduce the legal risk of processing data.

These technical measures and plans require enterprises to consider the protection of personal information and other rights and interests at the beginning of using face recognition technology, and protect the rights and interests of personal information by default design.