Should police body cameras have facial recognition tech? Axon, the largest U.S. maker of devices, says no
The country's largest maker of police body cameras said Thursday that it would not add a facial recognition feature to its cameras, a move that coincides with growing public opposition to the technology, including attempts by some cities to ban its use.
Axon announced its decision in response to a report published by its ethics advisory panel that urged the company not to pair its best-selling body cameras with software that could allow officers to identify people in real time based on their faces. The ethics board cited facial recognition's accuracy problems, which the board said could lead to false identifications, particularly of women and people with dark skin. The technology also could lead to expanded government surveillance and overzealous police activity, the board said. The board is made up of experts in the fields of artificial intelligence, computer science, privacy and civil liberties, as well as current and former police officials.
In a statement posted to Axon's website, CEO Rick Smith said current facial recognition technology "raises serious ethical concerns." But Smith also said that his team of artificial intelligence researchers would "continue to evaluate the state of facial recognition technologies," leaving open the possibility of adding the software to body cameras in the future.
The self-imposed moratorium was welcomed by civil rights and privacy advocates ─ but with skepticism. They noted that real-time facial recognition on police body cameras is not considered feasible at the moment, and they expressed concern that Axon could reverse course once that changed.
"This is ultimately an issue about the kind of of society we want to live in, not about technical specs," said Harlan Yu, executive director of Upturn, which monitors police agencies' body camera policies, and who has been an outspoken Axon critic.
Rather than rely on pledges from technology companies, lawmakers should impose regulations on how facial recognition is used, the advocates said.
That is already starting to happen at the local level.
Many police departments use facial recognition to match people caught on surveillance footage committing crimes to databases containing mugshots or driver's license photos. Those programs rely on still images taken from the footage. But companies are now developing systems that could allow police to identify people from live video footage, such as body cameras. Fears of runaway government surveillance have prompted recent attempts to curb facial recognition's growth.
Last month, San Francisco became the first city to ban local government use of facial recognition, with Oakland, California, and Somerville, Massachusetts, expected to enact similar restrictions soon. California's state Legislature is considering a bill that would ban the use of facial recognition on police body cameras.
"Axon leaves open the possibility that it may include face recognition in the future, which is why we need federal and state laws ─ like the current proposal in California ─ that would ban the use of facial recognition on body cameras altogether," said Jennifer Lynch, surveillance litigation director at the Electronic Frontier Foundation, a civil liberties nonprofit.
Smith, the Axon CEO, said the company will continue to sell programs that automatically detect the appearance of a face in a video and when a particular face appears multiple times. Those programs don't identify people; they make it easier for police to redact faces from videos before they are released to the public, he said.
James Burch, president of the National Police Foundation, which advises departments on the deployment of new technology, said it was smart for Axon to show "corporate responsibility" on a controversial issue. "How long it will last is a key question as market pressures and demands for more advanced capabilities increase," he said.
Axon has the largest market share of any body cam manufacturer in the United States; it says it supplies cameras to 47 of the 60 biggest police agencies. It does not say how many police agencies it has under contract, but says that more than 200,000 of its cameras are in use around the country.
Given the circumstances, Axon had little to lose from its announcement, at least in the short term.
Brendan Klare, CEO of Rank One Computing, whose facial recognition software is used by many police departments to identify people in still images, said he saw Axon's announcement as a way to make the company look good while making little substantive impact.
"The more important thing to point out here is that face recognition on body cameras really isn't technically feasible right now anyways," Klare said in an email.
A couple hours after Axon's announcement, the head of U.K.-based Digital Barriers ─ which is trying to break into the U.S. body camera market with its facial recognition-enabled devices ─ tweeted that Axon's move was good news for his company.