This article first appeared in the Planet Money newsletter. You can sign up here.
In August 2022, my truck was stolen on the streets of San Francisco. I had only lived in the city for a year, but it was the third time it was stolen or broken into. This led me to report on the epidemic of car crime around the city and how anti-theft technology made some cars harder to steal than others. The Bay Area then had one of the highest rates of car theft in the nation.
Four years later, something remarkable has happened: car thefts in SF are down almost 59%, and car break-ins are down more than 82%, according to our analysis of SFPD crime data comparing January through August of 2026 with the same months in 2022. And once again, technology is a prominent part of the story.
San Francisco's law enforcement leaders tell us that greater use of surveillance technology — including a controversial, AI-powered license plate camera system made by Flock Safety — has been crucial to this drop in car crime. How much credit does Flock actually deserve? And even if it does help reduce crime, can cities use this high-tech dragnet without trampling on everyone's constitutional rights?
As reports of officers abusing Flock have made headlines and the company has become increasingly controversial, hundreds of cities and counties have ended or suspended their contracts with the company, or decided not to renew them. But some, like Los Angeles and Denver, made a lot of hoopla about dropping Flock only to keep scanning license plates with a competitor surveillance system made by Axon — in a sense, ditching Coke for Pepsi.
San Francisco's leaders, however, remain committed to using the Flock system. Instead of getting rid of it, they're seeking to tighten the rules around its use and make misuse of its data a crime. Their approach could become a test case for cities around the country: can they use this sort of high-tech surveillance system to fight crime while putting the brakes on abuse of it?
Today in the Planet Money newsletter, we dig into the evidence and the debate over what these systems can do — and what they could cost us.
San Francisco gets tougher on crime
When my truck was stolen in 2022, San Francisco was in the midst of a revolution in its approach to crime. Weeks earlier, voters had recalled District Attorney Chesa Boudin over what many considered to be his lax approach to criminal prosecution. Then-Mayor London Breed appointed one of Boudin's biggest critics, Brooke Jenkins, to replace him. San Francisco voters have since elected her twice.
"I still remember four years ago when our city was, quite frankly, a public safety disaster," Jenkins tells us. "We were the headlines on Fox News and every other channel for our city being lawless and out of control."
Jenkins began taking a more aggressive approach to prosecuting people accused of crimes. Meanwhile, the SFPD beefed up its tactics to fight car thefts and break-ins. In 2023, for example, the police increased patrols in tourist areas known for car break-ins and used "bait cars" to catch and build cases against organized car burglars.
But, Jenkins says, "one of the most important factors" behind the drop in crime came when the city introduced new crime-fighting technology.
In March 2024, San Francisco voters approved Proposition E, which, among other provisions, made it easier for SFPD to use drones and high-tech surveillance cameras. That same month, the SFPD began installing Flock cameras in strategic locations. These "automated license plate readers," or ALPRs, take pictures of passing vehicles. Flock's AI software scans their license plates and identifies details like a car's make, color, body style, roof racks, and bumper stickers. It catalogs these images into a searchable database, letting investigators look up when and where a vehicle was spotted, even if it has no visible plates.
In a statement to Planet Money, a spokesperson for the SFPD said, "ALPR has been instrumental in driving down crime in San Francisco. These tools enable our officers to identify crime vehicles with precision, and ALPR is assisting in reducing crimes that were previously very challenging to solve, like auto burglaries."
Jenkins described how Flock cameras and drones can work together to catch suspected criminals. In one case the SFPD publicized, the department says Flock cameras spotted a moving vehicle linked to an auto burglary the previous day and alerted officers. Officers deployed a drone, which flew in from above and then began recording suspects in real time as they smashed windows and stole luggage out of a parked SUV. The police then moved in and arrested them.
They can do all this, Jenkins says, "without putting innocent people in danger with dangerous police chases." Meanwhile, the surveillance system creates an evidence trail that her office can use to prosecute suspected criminals.
"What the license plate reader cameras have absolutely done — and we do have proof of this — is give the police the ability to catch people who have engaged in crime," she says.
Jenkins also believes high-tech surveillance helps deter crime, although that's harder to prove. "People now know that we are a city that does have surveillance tools, does have drones, and they've decided [that] maybe the risk is not worth the potential benefit of them coming into San Francisco to commit certain crimes."
What does the research say?
Empirical research on the effect of these new, sprawling networks of AI-powered license plate readers on crime is still in its infancy.
There's a ton of studies on old-school surveillance technologies, including "closed-circuit television" security cameras. A 2019 meta-analysis of 76 studies spanning 40 years concluded that these cameras are "associated with a significant and modest decrease in crime." They found the "largest and most consistent effects" in parking lots, where crime fell by about 37% compared with control areas without cameras, suggesting that surveillance cameras can help reduce car break-ins and thefts, at least in those settings.
Research on earlier generations of license plate readers, however, found little evidence that they reduced crime. These plate readers were typically mounted on police cars. They scanned nearby license plates and alerted officers when they matched a list of vehicles reported stolen or flagged for other reasons. A Department of Justice review of the evidence on these old-school plate readers found they were "ineffective" at reducing crime, although at least one study found that they helped police recover stolen vehicles and make arrests. That said, these car-mounted plate readers are considerably different from the extensive, continuously operating networks of AI-powered cameras made by Flock and other companies.
The most ambitious study of Flock's system so far is a recent one by University of South Carolina crime researchers Ian Adams and Scott Mourtgos. They analyzed data from 216 law enforcement agencies that introduced Flock cameras, comparing crime trends with places that had not adopted Flock. They found that deploying Flock cameras was associated with an 11% drop in car thefts in the first year.
This is only one study, and it has yet to be peer-reviewed. It's also not a randomized controlled trial, the gold standard for finding the actual causal effect of something. Other factors, beyond Flock, could still help explain the decline.
In San Francisco, for instance, isolating Flock's contribution is especially difficult. There was a bunch of stuff going on that could reduce crime rates. The city, for example, got a new district attorney, the police increased patrols and started using drones, and the decline coincided with a broader drop in crime across American cities as the country emerged from the pandemic.
Adams and Mourtgos try to untangle these kinds of competing explanations using a method widely used in economics known as "difference-in-differences." The basic idea: compare trends in different places and look for differences that can be attributed to Flock. So if car thefts were falling in places without Flock by some amount, the cameras shouldn't get credit for that amount of the decline in places with Flock. It's the difference between the two trends that helps estimate Flock's true effect. Mourtgos says they also ran a series of "stress-tests" to verify that the Flock effect they find is real.
For example, Mourtgos says, they compared vehicle theft with other property crimes that the cameras are less likely to prevent, like shoplifting. Car theft fell more sharply, suggesting their finding isn't merely picking up a general decline in crime due to other reasons.
" I think it's very suggestive that they were responsible at least for a good portion of that 11%," Mourtgos says.
This study gives us a reason, beyond the assertions of law enforcement leaders, to believe the Flock system deserves some credit for San Francisco's drop in car thefts.
Adams and Mourtgos, however, didn't examine Flock's effect on break-ins, where thieves steal belongings from parked cars. These crimes have been much more common in San Francisco than thefts of the cars themselves.
Most newer cars now come standard with a technology that makes them harder to steal: an electronic immobilizer, which prevents an engine from being started without a special, electronically enhanced key. As I previously reported, immobilizers became increasingly common in cars around the turn of the millennium. My stolen 1999 Toyota Tacoma lacked this technology, which made it easier to steal. However, even if you have a newer car with an immobilizer, that doesn't prevent someone from smashing through its window and stealing your luggage.
Car break-ins have fallen even more sharply than car thefts in San Francisco. How much credit does Flock deserve for that, if any?
The drone footage example shows one way Flock cameras can help the police track down car burglars. But anecdotal cases, even with compelling videos, don't prove the cameras caused a citywide plunge in break-ins. And there may be reason to doubt that. These cameras photograph moving vehicles, typically at places like traffic intersections. A community-built map of reported Flock locations, while possibly incomplete, shows large gaps in residential areas, where many break-ins happen.
How would Flock stop someone from smashing a car window in the dead of night on a street where no camera is watching?
The SFPD's answer: it's helped them catch repeat offenders, and that can prevent a lot of break-ins.
"These tools help us identify organized crews of auto burglars," a SFPD spokesperson says. "There are a small number of suspects who commit the vast majority of these crimes. By using technology like ALPR, we're able to identify the crews and make arrests."
But even if Flock cameras do help reduce crime, as law enforcement officials adamantly claim and at least one preliminary study suggests is true for car thefts, an even bigger debate remains. Leaders and countless citizens in both political parties worry these surveillance systems are handing too much power to authorities to invade our privacy, trace our whereabouts, and stomp on our civil liberties.
Even if it helps fight crime, at what cost?
Earlier this year, the American Civil Liberties Union (ACLU) launched a campaign with an unusually funny name for the organization: "Get The Flock Out!"
The lover of puns who named and runs this campaign is Chad Marlow, a senior policy counsel at the ACLU. As some cities switch to surveillance systems made by competitors, he says, he's mulling another pun-infused campaign, " Kiss Your Axon Goodbye." He says he is still trying to figure out a pun for a campaign against Motorola.
Marlow has a long list of reasons to be wary of Flock and similar systems. He disputes that there's strong evidence they reduce crime. He also warns that relying on these tools can lead to mistakes that have real human consequences.
He recently testified at a US Senate hearing alongside Lindsey Isaacs, who spent 13 days in jail after investigators wrongly linked her vehicle to a fatal hit-and-run, relying in part on images of her vehicle captured by a Flock camera. The camera had photographed her car; investigators wrongly concluded it was the one involved in the crash.
But even if these systems work as intended, Marlow sees a big constitutional problem: they track people's movements even when there's no suspicion of wrongdoing.
True, he says, we've long had technologies like red-light cameras and speed cameras, but those are meant to record potential traffic violations. Flock cameras record passing vehicles whether or not their occupants are suspected of breaking the law.
Flock alone has more than 120,000 cameras around the nation. The police can search records from their own cameras and, when other agencies have agreed to share, they can search the records from cameras far beyond their jurisdiction. Marlow says around 75 percent of Flock law enforcement customers have opted to share their data with participating agencies across the nation, effectively creating a quasi-national database that authorities can use to track the movement of vehicles.
There have been many reported cases of police officers abusing access to the databases of Flock and other surveillance companies.
In August, The Washington Post published a story with a headline that sort of says it all, "How rogue officers turned a nationwide camera network into a tool for stalking." They reported how numerous police officers have improperly used license plate reader systems, including Flock's. "Authorities have charged or accused at least 50 law-enforcement officers of using license-plate readers for unauthorized purposes, including to stalk women without their knowledge or consent," they reported. Flock's system was involved in 46 of those cases.
In a statement to Planet Money, Flock Safety spokesperson Paris Lewbel says that searches of its system "leave an audit record showing who searched, when and why." Lewbel says that, by the end of 2026, the company's law enforcement customers will be required to use case codes for searches, with exceptions for emergencies, and a tool that flags unusual searches for agency administrators. The company also plans to automatically suspend a user's access when their search activity meets criteria for abnormal use. These stronger safeguards, Lewbel says, "are designed to make misuse easier to detect and stop while preserving officers' ability to investigate crimes." But much of the oversight still depends on city leaders and police departments policing their own officers.
