Flock Found More Stolen Cars Than Oakland Could Chase

Oakland installed Flock cameras to find stolen cars.

Then the cameras found them.

In six months, Oakland’s system generated nearly 189 million license-plate reads and 247,024 hotlist alerts. More than 210,000 of those alerts were tied to stolen plates or stolen vehicles.

That is not a typo. Oakland had so many stolen-plate and stolen-vehicle hits that OPD says the alert volume became “astronomical,” and the department could not properly respond while handling priority and emergency calls. OPD reported the numbers here.

That is the real story hiding inside the Flock debate: the cameras did not fail. They exposed a city that could detect more crime than it could absorb.

The Cameras Found More Crime Than Oakland Could Absorb

Flock is supposed to turn license-plate reads into real-time police action. A stolen plate hits the hotlist, officers get an alert, and the department can decide whether to respond.

But Oakland’s data shows what happens when the hit volume overwhelms the response system. Stolen-plate alerts were not occasional. They were the dominant signal: 187,120 alerts in six months.

That matters because stolen plates are often how stolen cars, robbery crews, retail theft crews, and sideshow participants move through the city without being identified.

The problem is that an alert still has to become a response. Someone has to triage it. Dispatch has to route it. Officers have to be available. The stop has to be safe and legally supportable. If the car flees, pursuit policy kicks in. If there is an arrest, the case still has to survive charging and court.

Oakland’s Flock system surfaced a huge amount of potential enforcement work. OPD’s own report suggests the department did not have the staffing, configuration, or operational pipeline to absorb it.

OPD Didn’t Turn Flock Off Because It Failed

The most revealing part of OPD’s report is not the huge read count. It is the operational response.

OPD did not say Flock could not find stolen plates. OPD said the alert volume was too high to respond to properly. Flock Safety flagged so many stolen plate and vehicle hits in Oakland that they had to temporarily disable the alert feature because an understaffed OPD couldn’t keep up.

That flips the story.

The problem is not that the camera network is useless. The problem is that the camera network may be useful in a way Oakland is not staffed, configured, or governed to use.

This is exactly the kind of public safety failure Bay Area politics loves to avoid.

Privacy advocates can say the cameras are dangerous. Police boosters can say the cameras work. Both can be true. The hard question is whether the city has a public safety pipeline capable of turning camera intelligence into safe, lawful, effective enforcement.

Oakland’s own documents suggest the answer is: not yet.

San Francisco Is Next

San Francisco is not Oakland. But it is running a similar experiment.

In 2024, San Francisco began installing 400 automated license-plate reader cameras at roughly 100 intersections, explicitly to disrupt organized retail theft, car break-ins, motor vehicle theft, sideshows, and other criminal activity. The city contracted with Flock Safety to install and maintain the cameras. SF announced the ALPR rollout here.

It later shared early results from the first 100 cameras, saying the technology had contributed to arrests in cases involving organized retail theft, carjacking, robbery, and sexual assault.

That is the pro-Flock case.

The cameras can help police find people and vehicles tied to serious crimes. They can make it harder for organized crews to move through the city invisibly. They can give investigators evidence that used to be scattered across private security footage, witness memories, and luck.

But the Oakland lesson should scare San Francisco too.

If Flock generates a firehose of alerts, the question becomes: who triages them? Which categories are actionable? Which get ignored? Which are disabled? How many hits go unresponded? How many turn into arrests? How many turn into recovered vehicles? How many are false positives? How many lead to stops that never produce a case?