Book review: Kashmir Hill’s 'Your Face Belongs to Us' is an exploration of privacy in the age of advanced facial recognition technology

The rise in facial recognition technology has been in lock-step with the loss of personal privacy as public and private places get populated with increasingly powerful CCTV cameras. (Image by Vecstock/Freepik)

We don’t think twice before sharing our photos and videos online. Facebook, Instagram, X (formerly Twitter), YouTube, and Flickr are just a handful of repositories where people have been sharing photos for years. This vast trove of data, paired with advanced facial recognition and AI, has given birth to a cutting-edge technology company: Clearview AI. The New York Times’ Kashmir Hill was the first journalist to break the story on Clearview AI in 2020. Calling her specialisation, the ‘dystopia beat,’ Hill gives us a ringside view of a technology that intersects with personal privacy and law enforcement in Your Face Belongs to Us: A Secretive Startup’s Quest to End Privacy as We Know It.

The book starts with a thrilling account of how Hill came across Clearview AI and its co-founder, Hoan Ton-That. Clearview AI is a technology company with a database of over 30 billion faces scraped from all over the internet that helps it identify people with an accuracy of over 96 per cent. Using just one photo (it doesn’t necessarily have to be a ‘facing the camera’ profile photo), Clearview AI can return many places on the internet where this person’s photos exist. On extrapolating this data, Clearview AI can also identify aspects such as locations, contacts of the person under check, and other personal data. The facial data has been collected without obtaining any permission from the people. While the company hoped to sell it to the business sector in its formative years, circumstances evolved to make US law enforcement agencies the primary buyers of this tech. The company was not known outside a limited circle despite its products being used by US law enforcement agencies till 2020.

Ton-That is an intriguing character. This Vietnamese-Australian tech prodigy decided to move from Australia to the US and got his start as a developer of Facebook quizzes. We see Ton-That’s transformation from a San Francisco liberal to a MAGA-hat-sporting Donald Trump supporter to becoming politically agnostic. One of the earliest versions of Clearview AI, called Smartcheckr, was first used at a Trump rally, and Ton-That used it to license this technology to the Hungarian government of Viktor Orban. According to the presentation pitched to Orban’s team, Ton-That promised that Smartcheckr was fine-tuned ‘to identify people that Orban considered enemies: pro-democracy activists who believed in ‘open borders’”.

While Ton-That is under the radar for two-thirds of the book, we notice how he doesn’t shy away from his creation in the last third. There have been many high-profile interviews of Ton-That since the NYT broke the Clearview AI story. It almost feels like, given the support he now has from the law enforcement agencies, the fear of his company being shut down by future regulations doesn’t seem to bother him in the slightest. Additionally, some of the backers of Clearview AI boast names such as Peter Thiel, one of Silicon Valley's most influential tech investors who was close to Trump at one point.

The San Francisco startup culture also played a background role in ensuring that Ton-That could carry on his research despite the initial setbacks, including an embarrassing profile in Valleyway, an erstwhile popular gossip site covering Silicon Valley. The rich and famous individuals in this part of the US also were the beta testers of Clearview AI during its earliest days. Hill chronicles how they abused the tech behind Clearview AI for their personal needs. An example is how an investor used Clearview’s tech to get the lowdown on the guy on a date with his daughter without their knowledge. Or how Hollywood actor Ashton Kutcher almost broke the embargo on sharing the details of Clearview which he was given to try out. Even the Madison Square Garden owners used Clearview AI tech not to allow entry to anyone working in a law firm with cases against the arena. Had this been allowed to continue, the profession of opposition research specialists could have taken a hit.

The book's central themes are equal parts a profile of Clearview AI, the implications of advanced facial recognition technology, and how its rampant usage without regulations raises many questions. Hill has also detailed how computer vision and facial recognition technology have evolved over the last few decades. The historical background helps put the implications of facial recognition technology in the proper context. It’s shocking to see how generations of biases against certain sections of society (women and people of colour, in particular) have been overlooked in the quest to put this technology into mainstream usage. There have been six cases of false arrests of people — all black — only because facial recognition software identified them as criminals. Hill details the shocking arrest of a black man, Robert Williams, in front of his family, followed by time spent inside the jail, all because Clearview AI returned his face as a match for a crime that took place in another state. In this case, the police department never bothered to do their standard investigation and entirely relied on Clearview's facial recognition technology.

The rise in facial recognition technology has been in lock-step with the loss of personal privacy as public and private places get populated with increasingly powerful CCTV cameras. With the help of some examples from China, Hill illustrates how these camera companies are advanced enough to do live facial recognition and assign social scores to people. Thankfully, this kind of evasive technology has not become a norm yet. One of the constant spiels from Clearview AI is that its technology has helped solve crimes. While that’s true to some extent, using its whole database to run checks on criminals is unethical. Hill gives enough cases in the book where things didn’t work out as promised. Additionally, there’s the question of investigation fatigue, as was seen in the case of the arrest of Robert Williams. It’s not far-fetched to say that this technology can be weaponised like all other technologies.

Clearview AI demonstrates how nimble startups can race ahead of established Big Tech companies. Hill notes how Big Tech has been surprisingly more responsible in the matter. Facebook, for instance, had developed prototype glasses that would call out the names of the person you were looking at through those glasses. It was never released to the public. (Sure, with the release of its latest Ray-Ban smart AR glasses, things may or may not remain the same.) Google also paused releasing advanced facial recognition out in the wild. Given the ruthless business principles of Big Tech companies in general, it was interesting to see them hold cutting-edge technology back.

While Hill does address many aspects of facial recognition and how Clearview AI is still planning to go ahead with its business plans, there are few instances, if any, of what can be done about it. According to Hill, some states in the US have banned the use of facial recognition technology. Clearview AI has been routinely fined in European countries like France and Italy, and its use violates the EU's General Data Protection Regulation (GDPR). But nothing much can be done outside of these instances to contain Clearview AI. Given the massive database of images, its accuracy ratings with face matches, and instances where it can be used for humanitarian purposes (identifying faces of dead soldiers in the Russia-Ukraine war to notify the soldier’s family, for instance), there seems to be no stopping Clearview AI’s ascent.

Ton-That is willing to "duke it out with critics, to fight lawsuits, to challenge governments that deemed it illegal, and to give as many interviews as needed to convince people that in the end their faces we not solely their own."

He concludes by telling Hill that it's time for the world to catch up, and it eventually will.