In this excerpt from “Your Face Belongs to Us” (Simon & Schuster, 2023), journalist Casimir Hill recalls the emergence of facial recognition technology company Clearview AI. Clearview AI has catapulted public awareness with its artificial intelligence (AI) software. You can identify almost anyone with just one shot of their face.
In November 2019, I had just started as a reporter for the New York Times and received information that was too outrageous to be true. A mysterious company called Clearview AI claimed it could identify almost anyone based on just a snapshot of their face.
I was in a hotel room in Switzerland when I received this email. I was just six months pregnant and had been planning to fly overseas for a while. I was tired after a long day, but seeing the email cheered me up. My sources unearthed a legal memo marked “Privileged and Confidential.” In it, Clearview’s lawyers said the company created the photos by collecting billions of photos from the public web, including social media sites such as Facebook, Instagram and LinkedIn. A revolutionary app.
If you give Clearview a photo of a random person on the street, it will spit out all the places on the internet where you’ve seen that person’s face, potentially revealing their name as well as other personal information about their life. There is. The company was marketing this superpower to police departments across the country, but they were trying to keep its existence a secret.
Not so long ago, automatic facial recognition was a dystopian technology that most people only associated with science fiction novels and movies like Minority Report. Engineers first tried to do it in the 1960s, trying to program early computers to match a person’s portrait to a larger database of human faces. In the early 2000s, police began experimenting with searching mugshot databases for the faces of unknown criminal suspects. However, the technology has proven largely disappointing. Its performance varied by race, gender, and age, and even the most advanced algorithms had difficulty doing something as simple as matching a facial photo to a grainy ATM surveillance still image.
Clearview claims to be different, touting a “98.6% accuracy rate” and a vast collection of photos unlike anything police have used before.
If this is true, it’s a big deal, I thought to myself as I reread the Clearview memo, which was never intended to be published. I’ve been covering privacy and its steady erosion for more than a decade. I often describe my beats as “the impending technological dystopia and how to avoid it,” but I’ve never seen such an audacious attack on anonymity before.
The term privacy is notoriously difficult to define, most famously described in an 1890 Harvard Law Review article as “the right to be left alone.” The two lawyers who wrote this article, Samuel D. Warren Jr. and Louis D. Brandeis, argued that the right to privacy should be protected by law, along with other rights such as life, liberty, and private property. he claimed. It is already enshrined. These were inspired by a novel technology at the time, the portable Eastman Kodak film camera invented in 1888, which allowed people to take their cameras outside the studio to take “instant” photos of everyday life. It has become. Nosy reporters.
Warren and Brandeis argue that “instantaneous photography and the newspaper business have invaded the sacred realms of private and domestic life, and numerous mechanical devices are now “what is whispered in the closet advertised from the roof of the house.” There is a risk that predictions such as ‘This will happen’ will come true.” ”
Related: Humanity faces a ‘catastrophic’ future if AI is not regulated, says ‘Godfather of AI’ Joshua Bengio
This article is one of the most famous legal essays ever written, and Louis Brandeis later joined the Supreme Court. But privacy wasn’t as deserving of protection as Warren and Brandeis said it was. More than a century later, there are still no comprehensive laws that guarantee Americans control over what photos are taken, what is written, or what happens to their personal data. Meanwhile, companies based in the United States and other countries with weak privacy laws are developing ever more powerful and invasive technologies.
Facial recognition has been of interest to me for a while. Throughout my career, I’ve covered major new products from multi-billion dollar companies in places like Forbes and Gizmodo. Apple and Google are allowing people to see and unlock their phones. Digital billboards from Microsoft and Intel equipped with cameras that detect age and gender to display relevant ads to passersby.
Concerns about facial recognition technology have been growing for decades. (Image credit: vasare/Getty Images)
I’ve written about how this sometimes clunky and error-prone technology excited law enforcement and industry while horrifying a privacy-minded public. Understanding what Clearview claims to be able to do reminded me of a federal workshop I attended in Washington, DC, many years ago. There, industry representatives, government officials and privacy advocates sat down to hammer out the rules of the road.
The only thing they all agreed on was that no one should publish an application to identify strangers. They said it was too dangerous. If some weirdo at the bar takes your picture, within seconds he might find out who your friends are and where they live. It could be used to identify anti-government protesters or women who walk into family planning clinics. It will become a weapon of harassment and intimidation. Accurate facial recognition on a scale of hundreds of millions or billions of people was the third rail of technology. And now Clearview, an unknown player in the space, claims to have built it.
I was skeptical. Start-up companies are notorious for making grandiose claims that actually turn out to be snake oil. Even Steve Jobs famously lied about the features of the original iPhone when he first unveiled it on stage in 2007. *
We tend to believe that computers have near-magical powers, that they can find a solution to any problem, and that given enough data, they can ultimately do it better than humans. As a result, investors, customers, and the general public can be fooled by outrageous claims and digital tricks by companies that want to do something great but are not quite there yet.
But in this confidential legal memo, Paul Clement, a prominent Clearview attorney who served as U.S. attorney general under President George W. I tried it and it says, “It returns fast and accurate search results.”
Clement said more than 200 law enforcement agencies already use the tool and that “when used for its intended purpose, Clearview does not violate the federal Constitution or any relevant existing state biometric or privacy laws.” I wrote that I had decided that. Not only were hundreds of police departments secretly using this technology, but the company hired high-powered lawyers to reassure police officers that they were not using it to commit crimes. Ta.
I returned to New York with my impending birth as a deadline. It took me three months to get to the bottom of this story, and the deeper I dug, the stranger it got…
Concerns about facial recognition have been growing for decades. And now this vague fool has finally found his form. It was a small company with a mysterious founder and an unfathomably large database. And none of the millions of people who make up that database agreed. Clearview AI represents our worst fears, but it also offers an opportunity to finally confront them.
*Steve Jobs solved the prototype iPhone’s memory problems and frequent problems by having his engineers spend countless hours finding a “golden path,” a specific set of tasks that the phone could perform without glitches. We’ve hidden the crash and figured out a faster way.
Kashmir Hill’s Your Face Belongs to Us: The Secretive Startup Dismantling Your Privacy has been shortlisted for the 2024 Royal Society Trivedi Science Book Prize, which recognizes the most popular scientific works from around the world.
