“I’d already got used to paying using Face ID. What do I do if I can’t remember my password?” “Bring back fingerprint scanning!” “The gates are fitted with facial recognition technology and didn’t open, so I couldn’t get into my own home!” These are just a few of the many thousands of indignant comments that Chinese people are currently posting on social media such as Weibo.

With World Health Organization officials demonstrating how to wear surgical masks and insisting that they really do offer protection from the coronavirus epidemic, many pharmacies and stores in China have sold out of them as Chinese people buy them in bulk. But this almost universal wearing of masks has had another unintended effect. It has essentially destroyed one of China’s technological achievements: its facial recognition system. 

Back in 2017, China already had over 20 million surveillance cameras equipped with face recognition technology. By the end of 2020, that number is expected to pass the 600 million mark. In theory, any high-resolution camera can be connected to the face recognition system. 

China is actively introducing the technology into various aspects of everyday life. Last year, about a thousand stores across the country started allowing customers to pay using Face ID, in which the customer simply looks at a special terminal, and the amount due is automatically deducted from their account. 

More than 100 million Chinese people have signed up for facial payment services by simply uploading photos of themselves to a mobile app. Today, nearly half the ATMs operated by China Construction Bank—one of the country’s four main banks—can perform operations using facial recognition technology. 

It can also be used to pay for public transport tickets. It isn’t just used for payment, either: the Beijing subway announced last fall that facial recognition would be used to classify passengers and determine who should undergo additional security checks when entering the subway.

At many tech companies, facial recognition technology is also used instead of electronic passes to open doors and grant access to members of staff. And last year, the Chinese authorities made it obligatory for anyone buying a cell phone SIM card to undergo a facial recognition procedure.

From time to time, this penetration of the technology into everyday life prompts accusations that the state is encroaching on basic rights to a private life and data confidentiality. Now, the coronavirus epidemic has revealed that a simple medical mask can paralyze the all-seeing Big Brother.

The first people to realize this were protesters in Hong Kong last summer. At first, they started donning glasses and surgical masks to protect themselves from the tear gas being used against them. But it turned out that the masks had the additional advantage of protecting wearers from being identified by police surveillance cameras. The Hong Kong authorities tried to ban the wearing of masks on the city’s streets, but the ban was declared unconstitutional by the High Court. 

Bans on wearing masks have existed in other countries since long before the invention of facial recognition technology. In the United States, for example, fifteen states banned the wearing of masks in the mid-twentieth century in an attempt to stamp out the activities of the Ku Klux Klan. The federal authorities adopted the same laws much later, amid the Occupy Wall Street protests. 

Russian laws governing protests also forbid participants from concealing their faces, including by wearing masks. In most countries, including Russia, the ban on masks only applies to protesters and participants of other large-scale events. In addition, other methods of outsmarting facial recognition technology, such as by applying face paint in a certain way, are a gray area. In any case, how can the law regulate the use of items such as surgical masks, which might become essential to some people’s health? 

The outbreak of coronavirus in China has exposed the weak spots of the country’s Big Brother system. It turns out that all the investment made in the public monitoring system can be rendered worthless by a simple mask. Now the Chinese leadership has no choice but to increase investment in improving its technology. It needs both to learn to recognize people in masks and to find solutions to more complex methods of disguise. 

That is, after all, possible. Panasonic facial recognition software using deep learning technology is able to identify people from images, even when up to 50 percent of their face is concealed—but it can only identify someone wearing a mask if their face is already registered in the software’s database. 

Chinese tech companies are also improving their facial recognition technology, as well as looking for alternative ways of identifying images of people, such as by the way they walk. Gait recognition technology has already been tested by police in places including Beijing and Shanghai, and has the advantage of being able to recognize people from up to 50 meters away, and even from behind. 

Now Chinese companies are also urgently trying to build algorithms that would recognize faces wearing masks. Two companies say they have created automatic access systems for office and residential buildings that work even when faces are largely concealed by surgical masks. For now, however, neither company has provided objective data that would make it possible to verify their claims or establish a margin of error.

The Chinese state is closely following the development of artificial intelligence. China has the highest concentration of “unicorns” in this field (start-ups with an estimated value of $1 billion and above). The most successful of them actively cooperate with the government and state companies.

Last November, China’s standardization administration brought together twenty-seven Chinese companies to develop a national standard for facial recognition technology. It’s not yet clear exactly what this working group will do. It’s possible that the weak spot exposed by the mass wearing of masks due to the coronavirus will be taken into account during the development of a unified standard. 

This publication is part of the Sino-Russian Entente project carried out with the support of the UK Foreign and Commonwealth Office. 

By:
  • Leonid Kovachich