Why the Clearview App Might Truly End Privacy as We Know It
The new app, used by law enforcement, scours every picture on the internet to find your face.
“now that I am about to leave this world, I realize there is nothing more astonishing than a human face” –John Ames in Marilynne Robinson’s novel Gilead
In Christopher Nolan’s The Dark Knight, Lucius Fox made the following comment about a sonar machine that non-consensually used people’s cell phones to find the Joker: “Beautiful. Unethical. Dangerous. You’ve turned every phone in the city into a microphone…”
Clearview, covered by the New York Times (article and podcast) as the app that “might end privacy as we know it,” can be explained with an almost identical description: “Beautiful. Unethical. Dangerous. You’ve turned every picture on the internet into a surveillance camera…”
Clearview, originally called “Smartcheckr,” is a facial recognition tool created by Hoan Ton-That. Given any picture of someone’s face, it looks for matches in public pictures from Facebook, YouTube, Twitter, Instagram, Venmo, news websites, and more. So why is it a threat to privacy as we know it?
Like the machine in The Dark Knight, Clearview operates at a massive scale. Batman’s machine and Ton-That’s machine both use sensors from cell phones — the former uses the sonar component, and the latter uses the camera. In both cases, the machine uses people’s property without their informed consent, and in both cases, the machine is used to find people.
The main divergence between Batman’s machine and Ton-That’s machine is in who gets to use it. In Nolan’s story, it’s the “good guys.” Actually, only one good guy: Lucius Fox says that “no one should have that kind of power,” and destroys the machine immediately after finishing the mission. Even a superhero was not allowed to keep such a powerful tool.
The main divergence between Batman’s machine and Ton-That’s machine is in who gets to use it.
But in the case of Clearview, law enforcement agencies throughout the country are actively using the app, and there appears to be no self-destruct button. In fact, the company is promoting the app with free 30-day trials.
Search or Surveillance?
Clearview’s homepage claims that they provide “search, not surveillance” because “analysts upload images from crime scenes and compare them to publicly available images.” They tout that their app “does not and cannot search any private or protected info.”
A couple examples from the New York Times report shows why the app does constitute surveillance and privacy invasions:
- A suspect’s face “appeared in the mirror of someone’s else gym photo”
- A demo search for the reporter’s face returned “photos of [her]self that [she] had never seen before”
If you are at a concert, on the subway, going for a walk in the neighborhood, or in some other public setting, and someone posts any picture with your face in it, you become visible to Clearview.
When a machine makes us visible in ways we do not know about, it diminishes our privacy. And a machine that gains such visibility without our knowledge engages in surveillance. This point could make up a whole separate blog post, but I will attempt to make it briefly: there is a false notion that a machine (such as Clearview) is not performing surveillance if nobody is actively watching you. Yet we know that security cameras do constitute surveillance even if nobody is actively watching you, because the ability to surveil is itself the surveillance. Surveillance is as much about infrastructure as it is about someone’s activity.
The ability to surveil is itself surveillance
Privacy or Security?
Clearview wisely promotes positive use cases on their homepage, noting that the app is “protecting the innocent,” “stopping criminals,” and solving “the toughest cold cases.” The “potential positive use” angle is important to grapple with, and I do not mean to dismiss it in just one blog post. However, I have two brief responses.
First, “absolute power corrupts absolutely.” I became familiar with this quote from Bruce Schneier’s great essay titled The Eternal Value of Privacy — I recommend it if you have time. In short, history teaches us that people who obtain an app like Clearview almost inevitably go on to use it maliciously. Power and corruption go hand in hand.
Second, we do not have to choose between safety and privacy. This thought is also in Schneier’s essay, and was popularized when Tim Cook publicly discussed Apple’s encryption policies in 2015 and beyond. The false binaries of “privacy vs. safety,” “privacy vs. national security,” and more, seem to be rampant in surveillance debates. But there is no such binary in the U.S. constitution, which supports common defense, general welfare, and individuals’ right to privacy.
We do not have to choose between safety and privacy
As a final point, the privacy discussion around Clearview is another case of AI acting as a synecdoche. “Put most bluntly, many people would not pick up a book about poverty policy in general — but are game to read a critique of the algorithms used to administer it,” a recent research paper observed. Here again, many people would not pick up a book about privacy in general, but with an AI app is involved, there may be more receptive ears.
