Facial Recognition Test Falsely Identifies 26 Lawmakers As Criminals

“One false match is too many," an ACLU lawyer said. The group backs a California bill banning use of the tests on police body cameras.

In an attempt to show lawmakers how dangerously inaccurate facial recognition technology can be in the hands of police, the American Civil Liberties Union decided to get personal.

On Tuesday, the ACLU’s Northern California branch released its findings from running photos of all 120 California state legislators against a database of 25,000 publicly available mugshots using common facial recognition software. The results were unsettling: the software identified 26 state legislators ― more than one in five ― as criminals. And a disproportionate number of those lawmakers were people of color

The ACLU supports California legislation to ban such technology from being used on police body cameras. Though no police departments in the state do so, the bill’s author, Assemblyman Phil Ting (D), wants to get ahead of the issue.

At a press conference with ACLU officials on Tuesday, he said the test involving the lawmaker faces was conducted “as a demonstration about how this software is absolutely not ready for primetime.”

He added, “While we can laugh about it as legislators, it’s no laughing matter if you are an individual who’s trying to get a job, if you’re an individual trying to get a home, if you get falsely accused of an arrest.”

California Assemblyman Phil Ting, the author of a bill to limit facial recognition technology's use in the state, was among s
California Assemblyman Phil Ting, the author of a bill to limit facial recognition technology's use in the state, was among several lawmakers falsely identified as criminals when the ACLU ran their photos against a mugshot database.

The technology’s use is already banned in San Francisco, the first major city to take such action.

Ting, a Chinese-American, was one of those falsely identified as a criminal in the ACLU’s test. So was Assemblyman Reggie Jones-Sawyer (D), who is Black and represents a minority-heavy district in Los Angeles. 

“Too often minorities are confused for others,” he said at Tuesday’s press conference. “I’ve heard of far too many cases of mistaken identity leading to arrests and, in the worst cases, death. This is without technology, which threatens to automate mistaken identity and risk the health and safety of countless people of color.”

He recounted a recent personal incident in which, because of a technical glitch in a police database, he was suspected by a California Highway Patrol officer of possibly driving a stolen car. If facial recognition technology had identified him as someone who’d committed a crime, Jones-Sawyer worried what would have happened.

“We would have a different situation right now,” he said, “and I would not be here right now.”

Matt Cagle, a lawyer with the ACLU, echoed Jones-Sawyer’s concerns.

“One false match is too many, especially when people’s lives are at stake or their freedom is at stake,” he said.

Ting’s proposal comes as officials in other states grapple with police already applying the technology. In Michigan, lawmakers are considering a bill that would place a five-year moratorium on its use after The Detroit Metro Times revealed that the municipal police department had been using a sophisticated facial recognition system for two years without city approval.

Ting’s bill was approved by the Senate Public Safety Committee and is expected to be soon debated on the Senate floor.