Recent Litigation Involving Clearview AI Leaves First Amendment Questions Unanswered

October 28, 2020

Recently, the attorney general of Vermont has brought a case challenging Clearview’s use of facial recognition technology and it has survived Clearview’s motion to dismiss. In their motion to dismiss, Clearview argued that it was protected under the First Amendment. It argues that Vermont’s action would violate its right to access public data on the internet.  Rather than examining this point in depth, the Vermont court stated that the “regulation sought by the state [was] content-neutral and, accordingly, subject to intermediate scrutiny.” This scrutiny gives strong deference to the states. The Vermont court essentially kicks the can down the road on whether or not companies such as Clearview are entitled to any First Amendment protections. The Vermont court quickly dismissed this argument and instead explains that the case has everything to do with how Clearview uses the information they gather, not on their ability to gather it. This argument carefully ignores Clearview’s contention and perhaps leaves open the possibility of future litigation on those grounds.

What is the enigmatic facial recognition software Clearview AI? In January 2019, the New York Times reported that over 600 law enforcement agencies used Clearview AI. Clearview’s customers are not limited to local law enforcement and is used by private companies as well. This company never intended to make itself a household name. In November 2019, a New York Times reporter found that the company’s website listed a “nonexistent Manhattan address as its place of business” and that it’s LinkedIn only showed one employee. That one employee’s name was fake.  Clearview advertises that as opposed to typical law enforcement databases which can only search for pictures such as mug shots or DMW photos, Clearview AI scrubs all of the pictures of an individual publicly available on the internet. When the same New York Times reporter tested the app on himself, Clearview returned pictures that he had never even seen before.

Across the country, police are using facial recognition technology to make after-the-fact arrests. This behavior has come to light after police departments have leveraged this technology to “track down and arrest individuals who allegedly participated in criminal activity during Black Lives Matter protests.” For instance, in August 2020, a woman in Miami was arrested for throwing rocks at police during a protest one month after the protest. Local investigators identified the woman using information available from her social media sites. These investigators also confirmed the Miami police leveraged Clearview AI. The Miami police department ensured the public that this technology would only be used to identify “violent protestors” and said that “a positive facial recognition search result alone does not constitute probable cause of arrest.” The balm that the police department provides to heal this breach of public trust is that if someone is peacefully protesting, then there is no way that the facial recognition information can be used against them. However, this statement indicates that the police do not intend to stop using this technology to gather information.

While the Vermont court’s decision to deny Clearview’s motion to dismiss seems like a victory for privacy, it raises First Amendment questions.

The issues posed by facial recognition software are felt in Europe as well. Clearview is under scrutiny from German regulators who question its right to scrub the internet for public information. Though Clearview has no customers in Germany, it has allegedly “copied billions of photos and turned the data into a ‘gigantic, easily searchable archive of facial images.’” Clearview defends its behavior saying that they are only compiling information that is publicly available. Clearview claims that Europe’s General Data Protection Regulation does not apply “to how the company processes data.”

Clearview’s behavior raises suspicion. It seems nefarious, dark, and shady. But these are intuitions, not legal analysis.  More states than Vermont, however, share this intuition as there are cases pending in state and federal court in California, Illinois, and North Dakota

While the Vermont court’s decision to deny Clearview’s motion to dismiss seems like a victory for privacy, it raises First Amendment questions. It raises the question of what, if anything, Clearview is doing wrong? Is Clearview right to say that the information they gather is public, and therefore the First Amendment protects their right to gather it? Is it possible for company to make this argument in the future? The Vermont superior court’s response to Clearview’s motion to dismiss does not answer this question.

Andrew Philip