Advertisement

LAPD bars use of third-party facial recognition systems, launches review after BuzzFeed inquiry

A LAPD officer wearing a face mask leaves police headquarters, where City Hall is reflected in a window.
Civil liberties advocates have questioned the efficacy of facial recognition software platforms that use images from outside the criminal justice system.
(Mel Melcon / Los Angeles Times)
Share

The Los Angeles Police Department has barred officers and detectives from using outside facial recognition platforms in their investigations after uncovering a handful of detectives had used a powerful commercial software platform known as Clearview AI without permission.

In a Nov. 13 directive sent to the entire agency, Deputy Chief John McMahon, who heads the LAPD’s Information Technology Bureau, noted that the only facial recognition system that LAPD officers are authorized to use is provided through the Los Angeles County Regional Identification System, which is maintained by the county and compares images input by officers against criminal booking photographs.

For the record:

6:10 p.m. Nov. 22, 2020A previous version of this story misspelled the last name of Clearview AI Chief Executive Hoan Ton-That.

Other platforms like Clearview, which compare images against millions of images posted on the internet, are not authorized for investigative use, he said.

Advertisement

“Department personnel shall not use third-party commercial facial recognition services or conduct facial recognition searches on behalf of outside agencies,” McMahon wrote. “Moreover, any department personnel using FRT shall attend the proper training and obtain a certificate of completion prior to using the system.”

Clearview has a history of courting officers to try out its system. The company’s chief executive, Hoan Ton-That, said in a statement that the LAPD “had a trial of Clearview AI as have many other law enforcement agencies around the country.”

Ton-That said the company’s software is used by more than 2,400 law enforcement agencies in the country and has features “to ensure responsible use.”

Civil liberties advocates have questioned the efficacy of facial recognition software platforms, particularly those like Clearview, which use images from outside the criminal justice system. Some critics and researchers have identified racial bias in facial recognition results.

LAPD Assistant Chief Horace Frank said the department began investigating the use of systems like Clearview by LAPD officers after it was contacted recently by BuzzFeed News, which said it had a list of more than two dozen LAPD officers who had purportedly used the outside software.

In an article published online Tuesday evening, BuzzFeed reported that documents it had reviewed “showed more than 25 LAPD employees had performed nearly 475 searches using Clearview AI as of earlier this year.”

Advertisement

As of Tuesday, Frank said the department had identified only two investigators who used Clearview AI on an investigation, though others appeared to have tinkered with the platform using non-investigative images. Some officers whose names were shared with the department by BuzzFeed denied ever using the Clearview platform, Frank said.

In the two instances in which LAPD officers did use the platform in investigations, images from a security camera were compared against the Clearview database, Frank said. He said he is aware of one arrest in a case in which the technology was used but did not know what role if any the facial recognition tool played in that arrest.

He said no arrests are made solely on the strength of a facial recognition match, and all require additional evidence.

Though officers have now been instructed not to use the Clearview system, those who did have not been punished, as they broke no rules with their prior usage, Frank said.

“The search complied with state law and our policy, as we didn’t have one forbidding third-party commercial facial recognition usage until last week,” he said.

Frank said the department is now conducting a more exhaustive review to determine whether officers have been using unsanctioned facial recognition platforms without the department’s knowledge. One LAPD source who spoke on condition of anonymity said detectives across the department are being asked whether they ever used outside technology, how they used it, whether it assisted in any investigations and, if so, in which cases.

Advertisement

The Times reported in September that the LAPD had used the county’s facial recognition software nearly 30,000 times since 2009, despite at times denying it ever used facial recognition. The Police Commission subsequently said it would review the city’s use of the technology and how it compared with other major cities.

The Times had specifically asked the LAPD in September whether officers were authorized to use Clearview AI software, sharing a screenshot of an email from a public records request that showed at least one detective in the Commercial Crimes Division discussing his use of the software with a Clearview representative.

At the time, LAPD spokesman Josh Rubenstein said that the officer identified in the email — Det. Joe Hopkins — was a California Organized Retail Crime Assn. coordinator and that Clearview had asked him to evaluate its product.

“He used an image that was not part of any criminal investigation for that evaluation, and received no results. This was a single evaluation Clearview requested and the system has not been accessed again by LAPD personnel,” Rubenstein said. “Additionally, the Clearview system has not been used for a criminal investigation by LAPD personnel.”

Advertisement