Clearview CEO argues for facial recognition regulation and education in live interview

Any tool as powerful as facial recognition needs to be regulated by the government, Clearview AI CEO Hoan Ton-That told an interviewer during a live online conversation Wednesday, despite the tremendous benefits it can deliver for law enforcement and other crucial social functions.

Ton-That joined The Washington Post for a live-streamed interview ambitiously titled: ‘The Path Forward: Facial Recognition Technology with Hoan Ton-That.’

The half-hour interview was conducted by Post reporter Drew Harwell, who asked about the ethics of scraping data from the public internet, the importance of how accurate facial recognition developed in the United States is relative to other countries, how Clearview ensures its facial recognition is not abused, and the investors behind the company.

Ton-That says the even though the images the company builds its database from may have been put online for a different purpose than the company is using them for, legitimate public interest makes the practice acceptable. He also says Clearview has declined to delete the images it collected from platforms which have since requested their deletion, accusing the company of violating their terms of service.

“I don’t think we want to live in a world where any big tech company can send a cease and desist and then control the public square,” Ton-That says.

People like academics researching bias reduction and misinformation need big datasets too, he notes.

To ensure ethical deployment, Clearview encourages customers to establish facial recognition policies, including oversight mechanisms and audit trails. While the company does not consider policing every biometric search by a customer to be its responsibility, Ton-That says it will revoke access from any client found to abuse its service, a capability which is an advantage of its cloud delivery model.

“We’re always thinking of ways to make this a safe technology in terms of deployment,” Ton-That said, and notes the company has never been reported to deliver a misidentification or prompt a wrongful arrest.

Improving controls remains one of Clearview’s top priorities, according to the CEO. Clearview is the first vendor in the space to require case number and type, he says as an example of its efforts in that regard.

When asked about investors, Ton-That notes that Clearview is “well-capitalized,” and that its investor mix has evolved as the company has grown.

While extremist provocateur Charles Johnson made some introductions to help in the company’s early days, Ton-That asserted that he had no active role, seat on the company’s board or other internal influence.

Harwell asks if Clearview’s volunteered support for Ukraine is an attempt to launder the company’s reputation, but Ton-That reframes the effort within the general mission to provide biometric recognition for public benefit. Like Clearview’s role in identifying U.S. Capitol rioters, Ton-That says these high-profile applications also help to educate the public on what it actually does.

Ton-That suggests one outcome of the use of Clearview’s facial recognition in Ukraine is the deterrence of war crimes by people who know they might be identified, and also suggests it could help with processing refugees more quickly and humanely. Clearview’s technology is now being used by 6 government agencies and 410 users in Ukraine, who have performed 14,809 searches as of mid-week, Ton-That revealed.

“My job and the job of the company is to continue educate people on how it’s actually used in practice,” he says in summary.

That education can also help inform the regulation which is necessary for any powerful technology, Ton-That says, comparing facial recognition to automobiles, which appeared on the market as a dangerous new technology before the introduction of layers of regulation.

The nature of that regulation was left for a future conversation.

Article Topics

biometrics  |  Clearview AI  |  ethics  |  facial recognition  |  Hoan Ton That  |  law enforcement  |  regulation