Facebook has agreed to pay a $550 million settlement over its use of facial recognition technology nearly a decade ago. This comes just days after a relatively unknown startup selling facial recognition systems to police departments caught the attention of Congress. It seems, after years of civil liberties advocates worrying, facial recognition technology is more powerful and more prevalent than ever. Neither a mammoth new settlement nor the piecemeal legislation nationwide seem suited to stop the takeover.
As it announced in the company’s quarterly earnings report on Wednesday, Facebook will shell out half a billion dollars to settle a 2015 class action lawsuit over its facial recognition software that suggested tags for people it identified in users’ photos.
What may have been a helpful feature for some also happened to be a possible violation of Illinois’ 2008 Biometric Information Privacy Act. The law — which happens to be the strongest biometric privacy law in the country — said that businesses must get permission before collecting biometric data. That essentially includes anything on your body that could be used to uniquely identify you, including scans of fingerprints, eyes, and faces. The class action suit claimed that Facebook’s photo-tagging feature, which it rolled out in 2010, did not do this.
Facebook claimed for years that it did nothing wrong. But after an attempt to throw the suit out was rejected by a federal appeals court in August 2019, Facebook discontinued the photo tag suggestions tool. Facebook’s appeal to the US Supreme Court was denied last week. Rather than risk a civil trial and a potential billion-dollar judgment, Facebook decided a $550 million settlement (which still has to be approved by a judge) was its best bet.
“We decided to pursue a settlement as it was in the best interest of our community and our shareholders to move past this matter,” a Facebook spokesperson told Recode.
Half a billion dollars is an incredible sum, but that quantity of money might amount to a mere inconvenience for Facebook. Given the company’s $21 billion revenue in its most recent quarterly earnings release, $550 million is pocket change. If, as our Peter Kafka has argued, the recent $5 billion fine from the Federal Trade Commission (FTC) for privacy failures won’t fundamentally change Facebook’s business model, a settlement worth a tenth of that won’t make a dent, either.
Of course, there are many other companies with much less money than Facebook doing much more with facial recognition technology. Clearview AI, as the New York Times and BuzzFeed News recently reported, claims to have scraped billions of photos from across the internet in order to build facial recognition technology that law enforcement agencies could use to identify suspected criminals. While Clearview claims to be working with dozens of police departments across the United States, the extent to which its technology is actually being used to fight crime remains unclear. Nevertheless, the company has brazenly pushed facial recognition technology out into the open in a way that bigger tech companies like Facebook have so far eschewed.
As a result, Clearview is facing its own class action lawsuit in Illinois, questions from a senator, and cease-and-desist letters from both Twitter and New Jersey’s attorney general.
However, it’s naive to assume that new, more aggressive forms of facial recognition technology will go away. China, Russia, and the United Kingdom use facial recognition for law enforcement on a massive scale. Just this week, Moscow rolled out a facial recognition system that’s been described as the world’s largest. A recent Recode investigation showed that the use of facial recognition technology is also increasingly prevalent in the United States.
It’s not just used to find suspected criminals and friends. Google and Amazon have their own facial recognition initiatives, and the technology has also been used for better, for worse, and for weird in recruiting, for unlocking phones, in dating apps, in retail, to get on trains, to find investors for your facial recognition company (a sort of facial recognition-ception), and for keeping track of pigs. This is despite public concerns over its implementation (as the reaction to Clearview AI shows) and data that shows it doesn’t work as well on non-white faces.
For what it’s worth, lawmakers are trying to regulate the use of facial recognition tech. Canada and the European Union have restrictions; in the United States, federal legislation has been proposed. In the absence of federal legislation, states like Texas and Illinois have issued bans on facial recognition.
Many other states have considered similar laws, and several cities including San Francisco and Oakland outlaw its use by the government. Evan Selinger, a professor of philosophy at the Rochester Institute of Technology who has written extensively about the dangers of facial recognition technology and has advocated for its ban, thinks such regulations are possible.
“The conversation has changed dramatically and political will is growing to stop the most dangerous surveillance technology ever invented,” Selinger told Recode. “Given the paper-thin divide between industry and the government and exposés like the one on Clearview, it is entirely possible that companies like Facebook will wind up restricted by bans, too.”
While Facebook did end the tag suggestions tool at the center of the class action lawsuit, the company didn’t stop its use facial recognition.
In 2017, it introduced a new “Face Recognition” feature with “the same technology we’ve used to suggest friends you may want to tag in photos or videos.” That tool replaced the beleaguered tag suggestions feature entirely two years later.
At this point, Facebook has been using facial recognition on our photos for almost a decade. Perhaps Selinger’s optimism is not misplaced. Or perhaps, in spite of Facebook’s very large but relatively insignificant fine, halting facial recognition’s use in the public or private sector is tantamount to putting the horse back in the barn (although facial recognition could probably make it easier to locate said horse).