Sometimes it’s nice to be recognized. Sometimes it’s not. But pretty soon almost wherever you go, it’s possible that someone will know who you are. Facial recognition technology is improving rapidly and spreading even faster. Its uses include mundane conveniences such as unlocking your smartphone or speeding your journey through passport control. But it’s also at the core of a vast system of surveillance China is imposing on its citizenry, most intensely on disfavored ethnic groups and dissidents. What China has shown is already possible is scaring even the companies working on what comes next. The voices calling for regulation include Microsoft Corp., which has warned against a “race to the bottom,” and Amazon.com Inc. But the measures privacy advocates are calling for go far beyond what tech companies seem willing to support.

The Situation

In May, the U.S. government said it was considering banning U.S. companies from exporting technology to a number of firms whose closed circuit cameras and facial recognition technology is used as part of China’s mass surveillance program. Also that month, San Francisco passed an outright ban on the use of facial recognition by city government, in one of the strictest regulations passed so far in the U.S. Police in the U.K. have deployed facial recognition software to find known pickpockets and shoplifters. In London, they faced criticism for fining a man who covered his face so it could not be scanned. In the U.S., police departments are writing their own rules. Law-enforcement officers in San Diego use software to match pictures of suspects to a database of mug shots. In Washington state, home to Microsoft and Amazon.com Inc., which sells a facial recognition product called Rekognition, Microsoft supported a bill that would have required prominent notices in public places where the technology was being used. Another bill, opposed by the company, would have prohibited its use by the government unless officials prove the technology would not lead to discrimination. Amazon opposed both measures, neither of which passed. Facebook is being sued over its photo tagging feature under an Illinois law requiring consent before biometric data is stored. Not all applications of the technology are controversial: In New Delhi, pictures of missing children are being checked against pictures of children found by police. Software vendors are selling banks on the technology for “know-your-customer” checks, while other systems are being used to register attendees at conferences.

The Background

Facial recognition technology was first developed in the mid-1960s with funding from U.S. intelligence agencies and the military. It advanced rapidly after 2012, as images from higher-quality cameras, vast new databases of faces, more powerful computers and a new class of algorithms made it possible to hone the artificial intelligence programs at the software’s core. In a study conducted in 2017, facial recognition programs outperformed all but the very best human experts. The software measures facial attributes — for instance, the distance between a person’s eyes or the angle of a chin — and turns that data into an individual template it can recognize even when the face is presented at different angles. The more advanced systems measure faces in three dimensions, meaning a photograph can’t fool them. In one infamous case in China, a prominent businesswoman was issued an automated ticket for jaywalking, when police cameras had actually captured her picture from an ad on the side of a bus.

The Argument

Proponents of the technology say it will make life more convenient and safer. Who wouldn’t want a system that could find wanted criminals or lost children? China cites the threat of terrorism to justify the intense surveillance of members of ethnic minorities in its western province of Xinjiang, where human rights officials say as many as 1 million Uighurs are now in internment camps. Other arguments for regulation stem from research finding that the systems are less accurate when processing the faces of members of minority groups, which has led to fears of unjustified arrests or harassment. Microsoft has responded by taking steps to make its programs more precise, while Amazon has disputed the research. International Business Machines Corp. developed a “Diversity in Faces Dataset” — only to face criticism that it had lifted images from sources such as Flickr photo-sharing accounts without obtaining permission. The tech giants have diverged in other ways, too. Microsoft says it turned down a request for its software from a law-enforcement agency that wanted to scan the faces of everyone its officers pulled over, while Amazon has pitched Rekognition to police agencies and federal immigration officials and Google says it’s chosen not to offer general purpose facial recognition programs until “important technology and policy questions” are resolved.

To contact the author of this QuickTake: Jeremy Kahn in London at jkahn21@bloomberg.net

To contact the editor responsible for this QuickTake: John O’Neil at joneil18@bloomberg.net

First published May 23, 2019

©2019 Bloomberg L.P.