How to protect our "face" in the age of face painting

Because of the disagreement to enter the zoo to brush the face, Zhejiang University of Technology special associate professor, Zhejiang University, Dr. Guo Bing...

Feb 25,2021 | SUE

Because of the disagreement to enter the zoo to brush the face, Zhejiang University of Technology special associate professor, Zhejiang University, Dr. Guo Bing, will Hangzhou Wildlife World to the court. The case became the "first case of face recognition" for domestic consumers suing businesses.

On 29 December 2020, the second trial of the case was held at the Hangzhou Intermediate Court, where Guo Bing filed a claim for invalidation of the content of the format clauses relating to fingerprint recognition and face recognition.

Not long ago, a video of "wearing a helmet to see a house" went viral, as staff at some sales offices would judge the identity of home buyers through face recognition and decide whether or not to give home purchase concessions.

In the Internet era, the development and application of new technologies such as "face recognition" has brought convenience, but also new challenges for the protection of personal information. In recent years, there have been many controversies surrounding face recognition. To a certain extent, these controversies reflect the widespread use of face recognition and the resulting public concern.

How can personal information be secured in the Internet era? What are the weaknesses of biometric authentication technology? Is there any regulation and supervision for biometric authentication application scenarios and technologies, including face painting? Science and Technology Daily interviewed relevant experts.

Biometric features are another ID card for us

At the airport and high-speed railway station to enter the station to verify identity, brush your face; to the hotel for check-in, brush your face; in the bank remote account opening, brush your face; online payment, through the access control, brush your face ......

In recent years, deep learning as the core of the rapid development of artificial intelligence technology, of which, the application of visual recognition technology landing quite widely, face recognition gradually from a few security scenes penetrate into people's daily lives. Since this year, the new crown pneumonia epidemic has brought about the demand for contactless scenarios, which has accelerated this process.

Face recognition is in fact a kind of biometric authentication, other biometric authentication there are fingerprint recognition, iris recognition, voice recognition, etc.. The most important feature of biometric authentication is uniqueness, for example, each person has a unique face, fingerprint and iris, etc.

Biometrics can therefore also be seen as another ID card for us. "Biometric authentication is the identification of personal characteristics, such as the identification of facial features, which is the same as checking the ID card number, which can refer to my personal identity." Li Qianmu, member of the party group and vice-chairman of the Jiangsu Provincial Science Association and director of the Information Division of Nanjing University of Technology, said.

Imagine an ID card that is either kept in your wallet or locked in a safe when not in use. If one day your biological ID card is stored by a property company, zoo, bank, hotel, etc. in a computer hard drive that you can't know about, will you still feel safe to swipe your face for authentication?

Pan Zhuting, chief strategy officer of Beijing Yongxin Zhicheng Technology Co Ltd, told reporters that passwords can be changed regularly and can be changed. But a person has only 10 fingerprints, two irises, two palm prints, one set of voice prints, and only one face. Biometric authentication is irrevocable and there is little remedy once its information has been leaked.

In a way, a face swipe that lacks security is no different from handing over your ID card and bank card PIN to random people.

"The promotion and application of such a technology should be fully justified in terms of its possible benefit to risk ratio. However, in the case of Hangzhou Wildlife World, we do not see the necessity and irreplaceability of using face recognition technology, nor do we see the full consideration and preparation of the risks by the user. Such actions to promote technology for the sake of promoting it are all cause for caution and reflection." Associate Professor Cheng Guobin of Southeast University thinks so.

Why face recognition has been repeatedly hacked

As early as 2017, during the 3-15 rights campaign, it was revealed that there was a security flaw in face logging: with a selfie of a viewer, the face authentication system of a mobile phone could be successfully "face swapped".

Since then, there have also been face data leaks, either because the database where the photos are stored was hacked or because staff sold copies of the data for profit.

Li Qianmu told reporters that there are two main weaknesses in biometric authentication: one is that biometric authentication mainly relies on images or videos for feature confirmation, and images and videos can be faked to some extent, "There is now a method called AI forgery, which is to 'make' a non-existent face through AI algorithms a non-existent face, or adaptively generate other people's faces, this algorithm is called GAN, also known as adversarial neural network, which can be trained by a large number of samples to generate some non-existent fake samples, pseudo-samples."

Secondly, biometric authentication is essentially character mapping, and in computers face features are described by numbers like 0 and 1. So even if the face does not exist in the system, by hacking the system, it is possible to input these features digitally and face recognition will potentially pass.

In October 2020, a research study involving more than 20,000 people - the Public Research Report on Face Recognition Applications (2020) - revealed that more than 90% of respondents had used face recognition, 60% believed that face recognition technology had a tendency to be misused, and 30% reported that they had already suffered loss of privacy or property due to face information being leaked and misused.

Compared to personal information leaks such as ID numbers and mobile phone numbers, there have not been many face data leaks that have come to the public's attention. However, there have been media reports that some online black marketers are using e-commerce platforms to sell illegally obtained face and other identity information and "photo activation" online tools and tutorials in bulk.

One positive change is that people's awareness of privacy protection has increased. According to the report, up to 80% of respondents expressed concern about whether and how raw face information is retained by the collector. In terms of the rules for handling face information, respondents were most interested in knowing "what technical and administrative measures are taken by the collector to ensure the security of the collected face information" and "whether face recognition technology is provided by a third party and, if so, who the third party is" and "in which scenarios face information is currently used and whether the purpose of use has changed".

Most of the experts interviewed said that they were happy to see the application and development of new technologies such as artificial intelligence, but that it was also urgent to strengthen the protection of personal information.

According to Li Qianmu, for the misuse of face recognition technology, one is to form a strong regulation to deter unlawful acts; the second is to carry out technical prevention, use artificial intelligence methods for anti-AI identification and discrimination, and establish a corresponding certification library or third-party certification centre to verify biometric authentication; the third is to speed up legislation and increase the punishment for illegal acts.

Looking around the world, some developed countries have already taken a step ahead in terms of legislation. in 2015, the United States released the report "Facial Recognition Technology - Commercial Use, Privacy Issues and the Federal Law Applicable to It", which imposes restrictions on the use of facial recognition technology by commercial entities to identify or track individuals. in 2018, the EU General Data Protection Regulation (GDPR) (GDPR) came into force, specifying that personal data is a data asset owned by individuals, which has been described as the "strictest" data protection bill in history.

China's legislation on the protection of personal data has also been accelerated. For example, the Civil Code includes biometric information of natural persons as personal information, and the Personal Information Protection Law (Draft) proposes to confiscate illegal income and impose fines for infringement of personal information rights.

It is undeniable that technology has strong instrumental properties and can be used by anyone. Some people use steel to build millions of buildings, while others use guns to slaughter thousands of living creatures. Therefore, Cheng Guobin believes that for the misuse of face recognition technology, in addition to legal regulation industry self-regulation, it must also be subject to profound ethical reflection and ethical governance.

"When developing a new technology, human beings always carry a certain value inclination or value pursuit. Making a consideration of the relationship between this specific initial value and the overall value that the technology should promote social development and human happiness is an important task for the ethics of science and technology." Cheng Guobin said.

In July 2019, the ninth meeting of the Central Committee for Comprehensively Deepening Reform considered and adopted the Plan for the Establishment of the National Ethics Committee for Science and Technology. The implementation of science and technology ethics review is to set the necessary ethical course and value bottom line for science and technology innovation, to clarify the ethical boundary of "something to do and something not to do" in science and technology activities, and to prevent the arbitrary opening of "Pandora's Box".

"However, as far as I know, China's ethical review mechanism is only relatively well developed in the medical sector, but is basically blank in universities and research institutes and enterprises. Compared with some developed countries in Europe and the US, there is still more room for improvement in terms of the coverage and completeness of the system for ethical review of science and technology in China." Cheng Guobin said the establishment of an institution is just the beginning, a whole set of mechanisms, policies, legal systems and a subtle awareness of scientific ethics need to be cultivated faster.

Website:

www.sundaymode.com

Print Mutual network Biosignature