Three-Year Moratorium on Facial-Recognition in Cop Cams Awaits Gov. Newsom’s Signature

State lawmakers have approved a bill that would temporarily ban police departments from using facial-recognition software in body cameras. If Gov. Gavin Newsom signs AB 1215 into existence, every law enforcement agency in California will be barred from using the technology for at least three years.

Civil rights activists backed the bill as a safeguard against a technology with the potential to violate people’s privacy. Proponents of the ban say it will prevent a tool meant to hold police accountable from turning into a means of spying on the public.

“Facial-recognition gives the government unprecedented power to track people’s location, identify them and to violate civil rights and liberties,” American Civil Liberties Union (ACLU) attorney Matt Cagle says. “They’re devices promised as police accountability tool—not as systems used for surveillance.”

Santa Clara County Supervisor Joe Simitian agrees.

The local lawmaker, who has positioned himself as a champion of civil liberties and privacy rights, says facial-recognition would undermine the purpose of police body cams. When he backed a measure to outfit local law enforcement with the cameras in 2014, he says the intent was to keep police in check—not to identify and track people.

“I did that for transparency and accountability reasons,” he says. “But I don't think anyone anticipated that body-worn camera would become a surveillance tool.”

Indeed, an ACLU survey of prospective voters in the 2020 presidential election found that 62 percent said they believe that body cams are designed to hold police accountable for their conduct. Only 9 percent said they consider them surveillance devices.

“Without my bill, facial-recognition technology essentially turns body cameras into a 24-hour surveillance tool,” Assemblyman Phil Ting (D-San Francisco) said in a statement.

Police unions, however, contend otherwise.

“The concept that it’s surveilling people in real time is not correct,” California Peace Officers’ Association spokesman Shaun Rundle says. “They make it seem like you’re out there surveilling people.”

Instead, Rundle explains that an officer’s body camera footage gets uploaded to a department server, which runs through a database of mugshots for potential matches. If there’s a hit, he says officers double-check to make sure it’s a true match.

Police departments conduct thorough due diligence, Rundle assures—a point that he says civil rights activists have overlooked. He also notes that most departments in California don’t even use facial-recognition. According to agency spokesman Sgt. Enrique Garcia, the San Jose Police Department certainly doesn’t.

Nonetheless, civil rights activists say the pending ban is an important step toward regulating a fast-evolving technology. “The threat that facial-recognition poses cannot be fixed by making it more precise,” Cagle says. “Even if it were perfectly accurate on body-worn cameras, it would be like requiring everyone to show their IDs to government.”

The bill initially prohibited police from using facial-recognition altogether. But lawmakers amended the bill from a straight ban to a three-year moratorium after police unions vehemently opposed the original language.

While Rundle acknowledges that facial-recognition is imperfect, he says an indefinite ban would have eliminated the chance for departments and manufacturers to improve the technology enough to use it as a way to catch suspected criminals.

Simitian says he realizes the potential public-safety benefit of facial-recognition. But the risks are too great to proceed with anything less than extreme caution.

“The issue is the wisdom with which we use that technology,” he clarifies. “You want to understand the imperfection of the technology and ask yourself: is there a way the technology can be helpful while still protecting people’s civil liberties? I don’t think we have wrestled that into the ground yet. That’s why I think the legislation is wise to say: hey let's figure this out before we go forward.”

While lawmakers wrestled with the implications of the technology on the state level, this county confronted the issue just a couple weeks ago.

Earlier this month, the Santa Clara County Sheriff’s Office brought its dash-cam surveillance policy up for review at a Board of Supervisors meeting. While Sheriff Laurie Smith’s underlings say the agency won’t integrate its dash-cam system with facial-recognition software, its policy doesn’t quite limit that possibility. It reads: “nothing in this Dash-Cam System Surveillance Use Policy shall preclude the use of static images obtained from dash-cam video with facial-recognition software.”

Simitian says the language alarmed him. After all, the ACLU conducted tests of facial-recognition technology and found that it can be unsettlingly inaccurate.

Though the ACLU’s methodology has been subject to debate, lawmakers found the test results disquieting. In testing the face-scanning software on 120 members of the California Legislature, 26 were misidentified as criminals—including Ting, who authored the facial-recognition moratorium. Worse still, the ACLU found the software more likely to misidentify minorities and women.

Clearly, there’s room for improvement. And Ting hopes the bill awaiting the governor’s signature will give regulators more time to address the technology’s failings.

“I want to make sure that we have some policies and procedures in place if we are going to use this imperfect technology and carefully curtail its use and eliminate the potential for misuse,” Simitian said from the dais at the Sept. 10 board of supes meeting.

That same day, the board referred the dash-cam policy to the Public Safety and Justice Committee for further review. What civil liberties advocates view as a threat to privacy, police see as a way to enforce the law. While both sides will no doubt continue to butt heads, the county’s stance is clear: facial-recognition is a double-edge sword with implications that merit scrupulous evaluation.

Should Newsom approve the ban on the body-cam spy-tech in question, that would at least give local officials more time to put the proper checks and balances in place.

Nicholas Chan is a journalist who covers politics, culture and current events in Silicon Valley. Follow him on Twitter at @nicholaschanhk.

2 Comments

  1. Anytime you have inaccurate results is disquieting. The results could be used as a backup to physical ID with drivers license. But never to be used as a pretense in shooting someone or mistreatment of individuals. Too many alarms going off with a flawed system.

Leave a Reply to sjoutsidethebubble Cancel reply