Talking Point

Facing the future

Chinese firms want to shape facial recognition standards around the world

China's Alipay adds beauty filters to face payments

Faced with the bill: an Alipay customer pays using facial recognition

When Minority Report came out in the summer of 2002, China was home to just 200 million mobile phone users and years away from launching its third-generation telecom network. Although we are still some way off replicating the dystopian backdrop to the Tom Cruise hit, some of the scenes imagined in the film are coming closer to reality – and more so in China than in other places, perhaps.

Product placement was pervasive in the Hollywood blockbuster as a symbol of intrusion into personal privacy. Today in Chinese stores ads can be sent directly to shoppers’ phones, once they have left their digital fingerprint through internet transactions or their faces are recognised by surveillance cameras.

Then there is the more controversial issue of predictive policing, provided in Minority Report by so-called ‘pre-cogs’, who could spot criminal activity in advance. In China, the detection work is falling increasingly on AI-powered computers that get smarter every day by working through a growing avalanche of data.

From tracking the purchases on phones to scanning peoples’ faces (and the way they walk), China’s surveillance efforts are also helping to create some of the world’s biggest tech unicorns, including the global leaders in facial recognition (FR) technology. This group has also become a sticking point in the trade and tech rows that have darkened Sino-US relations for more than a year. This is not only another tech battle between the world’s two biggest economies: instead as Chinese firms set out to shape global standards on a range of issues related to surveillance and data collection, more questions are being asked about the future of facial recognition around the world.

A golden year for China’s FR industry?

At the beginning of the year we wondered whether Megvii might be the FR unicorn “to watch over the next decade” (see WiC441). One of the so-called ‘four little AI dragons’, the eight year-old firm had just turned profitable and said it was planning to float its shares in Hong Kong. Three other facial recognition firms – SenseTime, Yitu and CloudWalk – were also receiving funding from powerful investors such as internet giant Alibaba (see WiC405). The financings put the quartet among the world’s most valuable FR firms.

Indeed, 2019 has proved another pivotal year for the Chinese FR industry. According to a three-year action plan published by the State Council two years ago, at least $1 billion of venture capital was to be dedicated to the sector before 2020. No later than 12 months from now, Chinese technology was also tasked with achieving a 97% “detection rate of facial recognition in complex dynamic scenes” and an “accurate recognition rate” that should exceed 90%.

That policy objective hinges on how the AI in facial recognition gets smarter and that is largely a function of the amount of data available. The good news for Chinese FR gurus is that this data is more abundantly available in their country than any other. Even so, the Chinese government is keen to collect more.

As just one example, the State Administration of Taxation contributed to the State Council‘s action plan by launching a new app on January 1 this year. It claimed the software was meant to simplify things for people applying for tax breaks. However, taxpayers have to have their faces swiped before they can fill out the forms online, raising some concerns about how the personal data was being captured (see WiC437).

How has China been pushing FR applications?

A similar face scan has just become mandatory for mobile phone users – coming into effect on December 1. The rule, which was first announced by the Ministry of Industry and Information Technology (MIIT) in September to fight fraud and “protect the legitimate rights and interest of citizens in cyberspace” – requires that the three telecom carriers scan the faces of all customers registering for new mobile phone services.

It is unclear how the law will apply to existing mobile accounts (there were 1.6 billion of those by the end of 2018). However, many commentators seem to believe that the same provisions will eventually be applied to the wider population of phone users. “The move brings with it considerable privacy and security concerns in one of the most tightly controlled online environments in the world,” CNN reported this week, adding that China already requires its citizens to link their social media accounts with their national ID cards, and that fuller implementation of the new procedure could remove any sense of anonymity online (up to 850 million people use their phones to access the internet in China).

While the new rule has stirred up the foreign press, it hasn’t stoked as much discussion in China. Maybe that’s because the public has already become more accustomed to FR identification, which is popularly applied in public spaces such as airports, as well as in more everyday environments like supermarkets and hotels. Some of the most popular apps, such as Alipay and Didi Chuxing, already require a face-scan authentication from customers. Cities have also been introducing FR in public services. In July, Xinhua said Beijing was in the process of installing facial recognition systems at the entrances of 59 public housing blocks, while other residents have their faces scanned when they throw out their rubbish, with ‘smart’ trash-cans keeping track of how much garbage they are throwing away. China’s capital has also joined about a dozen other cities in using FR on its subways and there is speculation that all of the country’s underground train networks could migrate to FR-enabled payment systems in the near future. Commuters would “smile and pay” through station gates instead of swiping their digital wallets.

Private sector firms have been responding to the government’s prompting by promoting FR as well. Take this year’s Singles’ Day, where online sales topped Rmb268 billion ($38 billion) – roughly five times what American retailers pulled in from online sales on Black Friday last week. Up to $1 billion worth of goods were sold in the first ‘bargain’ minute of Singles’ Day, with Jiemian reporting that about 60% of these ‘lightning fast’ transactions were settled with FR payment.

Does anyone mind having their face scanned?

The spread of FR into more areas of everyday life hasn’t been universally welcomed. One example is the education sector, where new technology is being applied in schools and colleges. Some institutions are relying on FR for safety purposes such as preventing strangers from entering school buildings. In other cases, schools are deploying it to prevent students from skipping classes or hiring cleverer surrogates to ace their exams. A growing number of high schools are also banking on FR to monitor student behaviour, with some surveillance equipment providers claiming it will detect whether their child is daydreaming.

However, some of these surveillance programmes have run into opposition (see WiC474 for a public relations disaster at a school using FR tech). And while parents worry about the personal privacy of their offspring, education experts have also pushed back, saying that the measures can undermine the learning process. A senior official from the Ministry of Education told ThePaper.cn in September that the central government was planning to “curb and regulate” the use of FR in schools. Yet there has been no further action from policymakers to date.

Facial recognition technology can present problems for regulators as well, like the face-swapping service Zao, which was China’s most downloaded app at one point in September. Developed by Nasdaq-listed dating platform Momo, it allowed its users to replace a celebrity’s face in movie scenes with their own. Partly concerned by potential disruption to the country’s FR database (as well as concerns over nefarious uses, including the spread of pornography), officials from the MIIT summoned Zao’s software bosses for a hasty meeting, demanding they “rectify and improve” their business model (see WiC465). Last week, the internet regulator also announced new rules governing the spread of ‘deepfake’ content, which uses AI to create realistic-looking videos where people can seem to say or do something they did not.

Contrary to the conventional belief that the Chinese are largely insensitive to privacy concerns, the state broadcaster CCTV noted last month that public awareness of the issue is actually on the rise. A case in point: a law professor in Zhejiang province has just sued a wildlife park for collecting his facial data as part of granting him admission. He refused, regarding it as a violation of his consumer rights. Widely dubbed as “China’s first face-swiping lawsuit”, CCTV said the case could go some way towards shaping future regulations for how FR is applied.

So a national standard on FR is needed?

Most of the problems related to FR applications arise from the absence of an overarching law regulating the new technology, Lao Dongyan, a law professor at Tsinghua University, told a symposium in Beijing last week.

“For most of the time, we don’t know our data is being collected and the storage and use of that data doesn’t follow legal requirements,” she said.

At the same conference, senior officials from MIIT also announced that 28 firms had formed a working group that would make recommendations on developing a national standard for FR deployment.

Both SenseTime and CloudWalk said they’ve joined the group. Other household names that are members include Ant Financial, Tencent, Ping An Insurance, Xiaomi and iFlytek.

It’s not clear if the recommendations that are made will eventually be turned into law, but citing insiders the Global Times said that the standards will address both technological and ethical issues, as well as easing any privacy concerns that the public may have.

Also importantly, the working group will be part of China’s broader attempt to shape more of the international standards for face recognition, which are still under discussion and study. “China is leading the world in this area and that’s why we have the responsibility to help set international standards,” an executive from the group told the Global Times.

How likely is it then that China will set global standards too?

Citing leaked documents, the Financial Times said Chinese tech firms including ZTE, Dahua and China Telecom are already proposing a “global standard” to the UN’s International Telecommunication Union (ITU) for FR and video monitoring.

Reportedly under discussion are recommendations for when FR can be deployed, such as surveillance sweeps by the police, by employers to monitor staff, and for spotting specific targets in crowds.

Human rights activists have already responded with alarm to commercial cases of deployment, including ‘smart streetlights’ which add video monitoring capabilities to lamp posts.

Regulations ratified at the ITU are commonly adopted by its 200 or so member states, particularly nations in Africa, the Middle East and Asia. Data collection from these markets is of particular interest to Chinese firms, which are looking to improve the accuracy of their FR algorithms, the FT said, particularly in identifying people of colour (CloudWalk is already working on a project with Zimbabwe’s government to use face-scans there to help to improve its capabilities).

Commercially, developing countries have often tended to go along with what is being put forward by the Chinese and the ITU, the FT added, as they don’t have the resources to set their own standards. Of course, that opens another route for the Chinese to export more of their own goods and technologies.

Previously, this quieter push into standards setting has gone largely unnoticed, according to Andrew Polk, a principal at market research house Trivium. “China’s government has sought to develop its own set of industrial standards for companies operating within its borders. That has made the effort mostly opaque to outsiders. Yet regulators are now starting to translate those standards into English – a clear sign that they’re meant to be exported overseas. And that should worry China’s competitors,” Polk warned in a Bloomberg op-ed last year.

The US government is now taking more notice, however. A number of Chinese AI firms were added to Washington’s black list earlier this year for alleged complicity in human rights violations, such as the massive surveillance programme said to be underway in Xinjiang. The ban has seen the likes of camera maker Hikvision cut off from the supply of American parts and the sanctions have also brought a sour end to the year for Megvii, which was planning to raise $1 billion through an initial public offering in Hong Kong. According to newspaper reports, it has just decided to postpone that IPO to next year, after Hong Kong regulators questioned the adequacy of risk disclosures it made relating to the impact of the US sanctions.

With that kind of political backdrop it’s going to be a lot harder for China’s facial recognition firms to make major sales in European and North American markets, not just on governmental concerns about national security, but also because of opposition from people who see FR as a threat to individual freedom.

Sure enough, news of the changes in the smartphone registration process soon stirred a response from activists outside China, including Evan Greer, the deputy director of an advocacy group demanding a ban on facial recognition for surveillance purposes. She told ABC News this week that Western nations shouldn’t just point the finger at China, because “the reality is that face surveillance programmes are spreading quickly in the US and Europe as well”.

However, her wider warning was that facial recognition technology is “ideal for authoritarian control” and that the new rules in China should be “a wake-up call to people everywhere who care about basic human liberty”.


© ChinTell Ltd. All rights reserved.

Exclusively sponsored by HSBC.

The Week in China website and the weekly magazine publications are owned and maintained by ChinTell Limited, Hong Kong. Neither HSBC nor any member of the HSBC group of companies ("HSBC") endorses the contents and/or is involved in selecting, creating or editing the contents of the Week in China website or the Week in China magazine. The views expressed in these publications are solely the views of ChinTell Limited and do not necessarily reflect the views or investment ideas of HSBC. No responsibility will therefore be assumed by HSBC for the contents of these publications or for the errors or omissions therein.