State Police admits AI use
Controversial facial recognition software aided investigations
After repeated denials, the Virginia State Police last week admitted that some of their detectives used a controversial facial recognition program during criminal investigations.
Last year, State Police told The Virginian-Pilot at least three times that it had never used Clearview AI — or any other facial recognition technology — and had no plans to. The Pilot also used the state’s public records laws to request emails sent between State Police employees and Clearview, which should have picked up messages the company typically sends to newly registered users. The agency turned over no such emails.
And in February of this year, a lawmaker said the governor’s top public safety adviser told her State Police were not using Clearview and never had. Del. Lashrecse Aird, D-Petersburg, said Brian Moran, the secretary of public safety and homeland security, gave her that assurance Feb. 9.
But it wasn’t true. On Friday, after another question from The Pilot, State Police spokeswoman Corinne Geller confirmed four investigators had used Clearview with free trial accounts for
five months, from September 2019 through February 2020. Two other investigators signed up for accounts but never used them.
Supervisors didn’t know investigators were using Clearview, Geller said. Only in February 2020 when one of them asked a supervisor about buying a license did it come to light. That supervisor discovered the six accounts and canceled them immediately.
Geller didn’t answer questions about how many investigations Clearview was used in or how many arrests — if any — State Police made as a result. She also didn’t answer a question about whether higher-ups had officially banned investigators from using Clearview or similar programs.
“I’m so floored,” Aird said Monday when The Pilot told her about the State Police admission. She said the fact that they “lied” about using Clearview makes her concerned about what else they could be hiding.
Through a spokeswoman, Moran said he learned on Monday that State Police agents had used Clearview, contrary to what he’d told Aird nearly two months earlier. That violates longstanding State Police policies not to use unapproved technologies, and on March 23, Superintendent Gary Settle told the agency’s roughly 2,100 sworn officers as much, Moran said.
Aird introduced legislation that she described as “a de facto ban” on local police using Clearview and other facial recognition programs. It passed the General Assembly unanimously and has been sent to Gov. Ralph Northam’s desk. He has till the end of Wednesday to sign it, veto it or propose amendments that would go back to the legislature for consideration.
But the legislation doesn’t cover State Police. Aird said she’s trying to see if she can amend her bill last minute so the ban would cover the agency of nearly 3,000 employees.
The State Police is one of more than 600 law enforcement agencies that started using Clearview between January 2019 and January 2020, according to The New York Times.
Clearview has aggressively marketed its app to law enforcement, promising it was “like Google Search for faces.” For police, the process is simple: Upload a photo of an unknown person or suspect to the company’s app, which then spits out public photos of those people, along with links to where they appeared. The system — powered by a database of more than 3 billion images that Clearview claims to have scraped from Facebook, YouTube, Venmo and millions of other websites — goes far beyond anything built by the United States government or Silicon Valley heavyweights, according to the New York Times report.
Though State Police investigators signed up for Clearview accounts, Geller said, the agency never paid for or officially OK’d using the facial recognition program, which is why none of the top brass knew Clearview was being used until later.
Because of the way Clearview offers trial accounts, that’s not unusual. Officers can start using the app with little more difficulty than downloading Facebook or Twitter.
That’s what happened with the Virginia Beach Police Department. Earlier this month, department officials admitted it also used Clearview, despite repeatedly telling The Pilot it hadn’t. The head of the detective bureau, Capt. Theresa Orr, said supervisors didn’t know officers were trying the app until someone asked a commander to pay to keep using it.
Supervisors said no, and now the police chief has ordered a department-wide review of how new technology is evaluated and adopted, especially when it has the potential to impact people’s privacy, police spokeswoman Officer Linda Kuehn has said.
The Norfolk Police Department is the only other Hampton Roads law enforcement agency that’s said it’s used Clearview, but stopped in February 2020 after two gang unit detectives tried it out for three months.
Even though all three agencies have said they stopped using Clearview, questions remain about the use of such technology to investigate crimes, whether in Virginia or across the country. There’s little oversight, legal or scientific, of how people’s images can be collected and used — even when they could help send people to prison.
Investigators in the nation’s roughly 18,000 law enforcement agencies are often able to unilaterally adopt whatever investigative tools they want, whether or not they’ve been rigorously tested by scientists or other law enforcement professionals.
And they often don’t have to tell anyone, even their own bosses, at least at first. In Norfolk, the mayor and a majority of City Council members said they didn’t know police had been using the technology.
“(That) really struck me as unacceptable,” Aird said.
Her bill, which she said was the result of The Pilot’s investigation into Norfolk police, would, among other things, require the General Assembly to pass a law specifically authorizing a police department to use the technology.