PROVIDENCE, R.I. (AP) 鈥� Facebook said it will and delete the faceprints of more than 1 billion people amid growing concerns about the technology and its misuse by governments, police and others.
鈥淭his change will represent one of the largest shifts in facial recognition usage in the technology鈥檚 history,鈥� Jerome Pesenti, vice president of artificial intelligence for , wrote in a blog post on Tuesday.
He said the company was trying to weigh the positive use cases for the technology 鈥渁gainst growing societal concerns, especially as regulators have yet to provide clear rules.鈥� The company in the coming weeks will delete 鈥渕ore than a billion people鈥檚 individual facial recognition templates," he said.
Facebook鈥檚 about-face follows a busy few weeks. On Thursday it announced its new name Meta for Facebook the company, but not the social network. The change, it said, will help it focus on building technology for what it envisions as the next iteration of the internet -- the 鈥渕etaverse.鈥�
The company is also facing perhaps its biggest public relations crisis to date after from whistleblower Frances Haugen showed that it has known about the harms its products cause and often did little or nothing to mitigate them.
Facebook didn鈥檛 immediately respond to questions about how people could verify that their image data was deleted, or what it would be doing with the underlying technology.
More than a third of Facebook鈥檚 daily active users have opted in to have their faces recognized by the social network鈥檚 system. That鈥檚 about 640 million people. Facebook introduced facial recognition more than a decade ago but gradually made it easier to opt out of the feature as it faced scrutiny from courts and regulators.
Facebook in 2019 stopped automatically recognizing people in photos and suggesting people 鈥渢ag" them, and instead of making that the default, asked users to choose if they wanted to use its facial recognition feature.
Facebook's decision to shut down its system 鈥渋s a good example of trying to make product decisions that are good for the user and the company,鈥� said Kristen Martin, a professor of technology ethics at the University of Notre Dame. She added that the move also demonstrates the power of public and regulatory pressure, since the face recognition system has been the subject of harsh criticism for over a decade.
Meta Platforms Inc., Facebook's parent company, appears to be looking at new forms of identifying people. Pesenti said Tuesday's announcement involves a 鈥渃ompany-wide move away from this kind of broad identification, and toward narrower forms of personal authentication."
鈥淔acial recognition can be particularly valuable when the technology operates privately on a person鈥檚 own devices," he wrote. 鈥淭his method of on-device facial recognition, requiring no communication of face data with an external server, is most commonly deployed today in the systems used to unlock smartphones."
Apple uses this kind of technology to power its .
Researchers and privacy activists have spent years raising questions about the tech industry's use of face-scanning software, citing studies that found it worked unevenly across boundaries of race, gender or age. One concern has been that the technology can incorrectly identify people with darker skin.
Another problem with face recognition is that in order to use it, companies have had to create unique faceprints of huge numbers of people 鈥� often without their consent and in ways that can be used to fuel systems that track people, said Nathan Wessler of the American Civil Liberties Union, which has fought Facebook and other companies over their use of the technology.
鈥淭his is a tremendously significant recognition that this technology is inherently dangerous,鈥� he said.
Facebook found itself on the other end of the debate last year when it demanded that facial recognition startup ClearviewAI, which works with police, stop harvesting Facebook and Instagram user images to identify the people in them.
Concerns also have grown because of increasing awareness of the Chinese government鈥檚 extensive video surveillance system, especially as it鈥檚 been employed in a region home to one of China鈥檚 largely Muslim ethnic minority populations.
Facebook鈥檚 huge repository of images shared by users helped make it a powerhouse for improvements in computer vision, a branch of artificial intelligence. Now many of those research teams have been refocused on Meta鈥檚 ambitions for augmented reality technology, in which the company envisions future users strapping on goggles to experience a blend of virtual and physical worlds. Those technologies, in turn, could pose new concerns about how people鈥檚 biometric data is collected and tracked.
Meta鈥檚 newly wary approach to facial recognition follows decisions by other U.S. tech giants such as Amazon, Microsoft and IBM last year to end or pause their sales of facial recognition software to police, citing concerns about false identifications and amid a broader U.S. reckoning over policing and racial injustice.
At least seven U.S. states and nearly two dozen cities have limited government use of the technology amid fears over civil rights violations, racial bias and invasion of privacy.
President Joe Biden鈥檚 science and technology office in October launched a fact-finding mission to look at facial recognition and other biometric tools used to identify people or assess their emotional or mental states and character. European regulators and lawmakers have also taken steps toward blocking law enforcement from scanning facial features in public spaces.
Facebook鈥檚 face-scanning practices also contributed to the $5 billion fine and privacy restrictions the Federal Trade Commission imposed on the company in 2019. Facebook鈥檚 settlement with the FTC included a promise to require 鈥渃lear and conspicuous鈥� notice before people鈥檚 photos and videos were subjected to facial recognition technology.
And the company earlier this year agreed to pay $650 million to settle a 2015 lawsuit alleging it violated an Illinois privacy law when it used photo-tagging without users鈥� permission.
鈥淚t is a big deal, it鈥檚 a big shift but it鈥檚 also far, far too late,鈥� said John Davisson, senior counsel at the Electronic Privacy Information Center. EPIC filed its first complaint with the FTC against Facebook鈥檚 facial recognition service in 2011, the year after .
___
Ortutay reported from Oakland, Calif.
Matt O'brien And Barbara Ortutay, The Associated Press
