Regulatory dust settles around Clearview, leaving one big facial recognition market

Regulatory changes and court decisions around the world leave Clearview AI with a limited, but vast target market for its facial recognition: American law enforcement.
CEO Hoan Ton-That tells MLex Market Insight that the company has reached a “turning point” in its fight against litigation. He also believes the company has moved beyond misconceptions among the media about how facial recognition is used.
The company has emerged from a series of disputes in court with what Ton-That sees as a solid position. A proposed settlement in Illinois “is a really big deal” for Clearview, he says, while a Vermont judge denied a request from the state’s Attorney General for partial summary judgement in December.
The judge cited Clearview’s submission that its assistance of investigations into the January 6, 2021 Capitol riot, Ukraine’s war effort and in defense of the innocent, saying: “The court cannot say that any reasonable Vermonter would find that the privacy violations ensuing from Clearview’s product outweigh the benefits.”
Ton-That does not see Clearview entering the European market. He cites the AI Act and GDPR as evidence that “Europe is taking a different view on not just data privacy, but AI and technology in general.” Neither is Clearview planning to set up operations in the UK, Australia, Canada, or other countries it has faced legal or regulatory pushback in.
Australian police supplied forensic material to an international database that was analyzed by U.S. law enforcement with Clearview’s facial recognition last year as part of Operation Renewed Hope. This has prompted fears Down Under that Crikey reports police are reticent to even talk about with the press.
In the U.S., dealing exclusively with law enforcement puts Clearview outside of the jurisdiction of most of the state laws recently passed.
“We’ve been growing quite quickly, especially in our state and local business now, in the US,” Ton-That says. “The federal business is really growing as well. It’s also at an inflection point where a lot of these US federal government customers have gone through their privacy policies, training policies around facial recognition.”
Clearview faces ongoing legal challenges in the U.S., including a call for California authorities to investigate its data-collection practices. Ton-That believes that Clearview is on solid ground in that regard. The company scraped data from platforms without creating accounts, which would have obliged it to follow the platforms’ terms of service.
An Israeli company called Bright Data was recently found by a U.S. District Court judge to have acted legally when scraping data from Meta’s Facebook and Instagram because it was not logged in.
Testing for accuracy
Misconceptions in the media may be less of a problem for Clearview than in the past, but they are still easy to find.
Benton County Sheriff’s Office in Washington held a public hearing to discuss a potential contract with Clearview last week, local CBS affiliate KEPR reports.
Washington Defender Association Felony Resource Attorney Sheri M. Oertel claimed that the company’s facial recognition algorithm has not been tested for accuracy or racial disparities. The claim is unchallenged by KEPR, despite having submitted algorithms to NIST’s ongoing biometric vendor test series since 2021. Ton-That specifically told Biometric Update in an email at the time that the company first submitted its technology to the 1:1 assessment, rather than the 1:N application it is normally used for, in order to have its demographic differentials evaluated.
An assessment of the bias of an algorithm submitted by Clearview in December, 2023 can be viewed on the NIST website.
Comments by Oertel reported by the Tri-City Herald, however, clarify what she means about the testing being inadequate: “They haven’t gone out and checked low quality, low light video surveillance that you get from a convenience store. They haven’t checked partial faces,” she said.
Article Topics
biometrics | Clearview AI | facial recognition | police | United States | Vermont
Comments