Thursday, October 15, 2026

SF Car Crime Plunged. How Much Credit Does Flock Get?

Valyrian News Network — — 7 min read

SF Car Crime Plunged. How Much Credit Does Flock Get?

Car crime in San Francisco has fallen sharply over the past four years, and city officials are crediting a network of AI-powered license plate readers for much of the improvement. But a close look at the data — and at the company behind the cameras — suggests the story is far more complicated than a single technology success.

According to NPR’s Planet Money, motor vehicle thefts in the city dropped almost 59% between January and August of 2026 compared with the same period in 2022, while car break-ins fell more than 82%. The steep decline coincided with a new district attorney, expanded patrols targeting tourist corridors, new drone powers, and the installation of Automated License Plate Readers (ALPRs) made by Flock Safety — a company whose cameras now watch more than 120,000 locations across the country.

A City That Changed Course

The turnaround followed a political inflection point. In June 2022, San Francisco voters recalled District Attorney Chesa Boudin amid widespread frustration over property crime and open-air drug markets. Then-Mayor London Breed appointed Brooke Jenkins, a vocal Boudin critic, as district attorney. Jenkins has since been elected twice and has become one of the technology’s most prominent defenders.

“I still remember four years ago when our city was, quite frankly, a public safety disaster,” Jenkins told NPR. “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.”

An SFPD spokesperson described the technology as “instrumental in driving down crime,” adding that it helps officers “identify organized crews of auto burglars” who account for a disproportionate share of offenses.

The cameras arrived in force in March 2024, the same month voters approved Proposition E, a ballot measure that eased restrictions on police use of drones and surveillance cameras. Flock’s system uses artificial intelligence to catalog not just license plates but vehicle details — make, color, body style, roof racks, bumper stickers — into a searchable database, capturing cars even when no plate is visible.

The Evidence Behind the Claim

The strongest study cited in Flock’s favor is a working paper by University of South Carolina researchers Scott Mourtgos and Ian Adams. It examined 216 agencies that adopted Flock cameras and compared them with non-adopters, finding an 11% decline in car thefts in the first year of deployment. The paper has not been peer-reviewed and is not a randomized controlled trial.

“I think it’s very suggestive that they were responsible at least for a good portion of that 11%,” Mourtgos told NPR. Even so, the finding carries a wide 95% confidence interval, ranging from a 17.3% drop to just 4.2% — and under equal- or population-weighting, the effect shrinks and is no longer statistically distinguishable from zero. The study also examined only car thefts, not the break-ins that dominate San Francisco’s car-crime problem.

When Flock promoted the study in August, trade outlet The Auto Wire argued the company left out critical caveats: the confidence intervals, the weighting sensitivity, a pre-deployment rise in clearance rates that undercuts one of its claims, a null result on vehicle recovery speed, and the paper’s own section on privacy and documented misuse. The distortion, the outlet wrote, was “one of omission rather than fabrication.”

Isolating Flock’s specific contribution is made harder still by a national crime decline as the U.S. emerged from the pandemic, alongside San Francisco’s concurrent changes in policing and prosecution.

Privacy Costs Come Into Focus

The same technology that helped drive arrests has also produced a record of misuse. In September 2025, The San Francisco Standard reported that SFPD had allowed out-of-state agencies to run more than 1.6 million searches of its database — a practice California law prohibits — including at least 19 marked as related to ICE. A June 2026 internal audit found 299 improper searches by analysts at a California law enforcement partner acting on behalf of federal and out-of-state agencies.

Nationally, The Washington Post reported in August 2026 that at least 50 officers had been accused of using license plate readers for unauthorized purposes such as stalking, 46 of them involving Flock. One Wisconsin officer conducted 179 database searches targeting an ex-girlfriend. Flock also drew scrutiny after privacy researchers found dozens of its cameras exposed on the open internet and after CEO Garrett Langley sent clients an email attacking activist groups as “terrorists.”

“It can be used to go after any vulnerable group or any group that the government agency that controls the data or can access the data wants to go after,” Chad Marlow, a senior policy counsel at the ACLU, told NPR. “We’ve seen it go after immigrant populations, against people seeking abortion care, against people who go to a gun show or a gun range, against people who are involved in political protests.”

By March 2026, at least 30 localities had canceled contracts or deactivated ALPRs, and hundreds more had ended or declined to renew Flock deals — though some, including Los Angeles and Denver, simply switched to competing systems from Axon.

A Test Case for Regulation

Rather than retreat, San Francisco is trying to thread the needle. On September 29, Mayor Daniel Lurie unveiled new privacy protections that cut SFPD’s access to ALPR data for investigative searches from 12 months to 30 days, require that every search be tied to a specific incident number with a documented reason, and mandate monthly reviews and independent audits by the Department of Police Accountability.

“Every active search will have to be connected to a specific incident number,” Lurie said in remarks reported by KQED. “The officer will have to state the reason for the search and the system will record who conducted it.” The mayor framed the balance bluntly: “This technology keeps San Franciscans safe, and we will not go back to a time where people can break into cars, rob businesses or commit serious crime and expect to get away with it.”

Jenkins offered a candid political rationale for the safeguards. “I was motivated by the fact that I didn’t wanna see us lose the opportunity to use this technology,” she said. “And if we don’t put the correct safeguards in place, then the people are going to say, ‘Then you shouldn’t be allowed to use it.’”

State lawmakers are moving in parallel. Since California does not currently make it a crime for police to misuse ALPR data, San José leaders want to change that, and Assemblymember Matt Haney is co-sponsoring legislation to impose criminal penalties. “You don’t get to use that information to track someone for personal reasons, share it with a friend or family member, or use it for anything unrelated to your job,” Haney told KQED.

Flock says it is tightening its own controls. A company spokesperson told NPR that searches “leave an audit record showing who searched, when and why,” that customers must use case codes by the end of 2026, and that a tool flags unusual searches and auto-suspends access for abnormal behavior.

What to Watch

The debate reflects a broader constitutional question that courts have yet to resolve: whether cities can operate warrantless mass surveillance of people not suspected of wrongdoing. Unlike red-light cameras, which capture specific violations, license plate readers record every passing vehicle.

For San Francisco, the bet is that transparency and strict access rules can preserve the crime-fighting benefit while containing the abuse. Civil liberties groups like SIREN argue the problem is structural, not fixable by policy. “Mass surveillance goes beyond one company, one vendor, and one ‘bad actor,’” organizer Kimberly Woo told KQED.

The coming months will test whether the city’s dual-track strategy — keep the cameras, criminalize the misuse — holds up. The success or failure of that experiment will likely shape how dozens of other cities, weighing their own crime data against their own privacy fears, decide whether Flock stays on their streets.