You’re at a gathering. A stranger snaps a photo of you with a smartphone, then proceeds to tell you your name, your employer, where you live and perhaps your political affiliation. If this party trick sounds like the stuff of science fiction, it’s not.
Face-recognition technology is nothing new. It’s widely used to unlock phones and other devices, and in a growing number of settings involving identity authentication, like airport security. You voluntarily register an image of your face and you get access to a service. But if you’ve ever uploaded a photo of yourself to a public social media site, chances are some companies are now using it without your consent.
Kashmir Hill, a tech reporter with the New York Times, begins her exploration of this disturbing trend at a hotel in Switzerland, where she receives an email from a source who has attached a legal memo from a mysterious company based in New York called Clearview AI. It has created a revolutionary app that can identify random people on the street. It can do this because it has scraped billions of photos from the web, including from sites like Facebook, YouTube, Venmo, Instagram and LinkedIn. Clearview has already been selling the technology to police departments but trying to keep it under wraps.
Back in the United States, Hill tries finding out anything she can about Clearview, but meets dead ends – fake addresses and sources who suddenly go cold. Finally, she doorstops an investor, David Scalzo of Kirenaga Partners, who opens up about it, gushing, ‘I’ve come to the conclusion that because information constantly increases, there’s never going to be privacy.’
This cynicism doesn’t sit well with Hill, who soon realises that Clearview has built a powerful surveillance weapon, one that could be used by authoritarian governments to identify and crack down on dissidents. She sets about unearthing as much information as she can about Clearview.
The startup was launched in 2017 by Hoan Ton-That, a Vietnamese-Australian software engineer who had created apps for Facebook, and Richard Schwartz, a former aide to Rudy Giuliani. Ton-That took an early interest in Donald Trump’s MAGA movement and the two were introduced by Charles C Johnson, an alt-right provocateur with powerful acquaintances, such as Peter Thiel, the billionaire conservative activist. Johnson and Ton-That had bonded at the 2016 Republican convention, where Johnson asked Ton-That whether he wanted to do something more than simply create Facebook apps. ‘This is going to be a lot of fun,’ Ton-That grinned. ‘Of course I’m in.’
To her credit, Hill contextualises Clearview’s activities by considering the long history of research into faces. This encompasses not only early attempts to train artificial intelligence systems to recognise human faces, but also the earlier pseudoscience of researchers like the Victorian social Darwinist Francis Galton, a believer in eugenics and phrenology. Johnson was eager to use modern computing tools to revive the related field of physiognomy, according to Hill.
Although Ton-That would later disavow Clearview’s hard-right connections, the fledgling business benefited from Johnson’s introductions and Thiel’s seed money. Clearview’s technology was marketed to security experts and law-enforcement agencies, and they were astonished at how effective it could be. Local police departments, the FBI and the Department of Homeland Security began using the system, which can provide not only a person’s identity based on a face photo but also links to sites where other public photos of the individual appear. Not only that, it can be used with augmented-reality glasses, allowing users to identify nearly anyone they see.
‘The weaponization possibilities of this are endless,’ Eric Goldman, co-director of the High Tech Law Institute at Santa Clara University, said in an exposé published by Hill in early 2020. By then, more than six hundred law-enforcement agencies had begun using the service, according to Clearview, all in secret. Facebook and Google had also developed similar technologies but refused to release them, deciding they were too dangerous to be made widely available. A case in point: in 2020, a Michigan resident, Robert Williams, was arrested and detained for nearly thirty hours on a charge of shoplifting because face-recognition technology erroneously matched an image of his face to that of the suspect.
After the tech giants told Clearview to stop scraping images, and lawsuits were filed in Illinois and Virginia, Clearview published a code of conduct stating that its technology was only available to law-enforcement organisations. But Hill showed this to be false. Individuals were using the app too. Grocery billionaire John Catsimatidis secretly did a background check on his daughter’s new boyfriend after snapping a photo of him. The magnate also used it to identify shoplifters in one of his stores.
Hill’s book is a breezy, compelling dive into the alarming use of face matching and the enormous consequences for privacy and civil liberties. It’s an engrossing cautionary tale about not just civil society trying to keep up with technology but also the dangers of posting personal and biometric information on the internet. As for Clearview, numerous privacy watchdogs in North America and Europe have declared illegal its use of public photos to identify people, but the company continues to generate interest and business: in July 2021 it closed a $30 million investment round, bringing its valuation to $130 million. Our faces are now public commodities generating private profit with little democratic oversight.