In the summer of 2024, a police officer in the Sacramento suburb of Roseville, California, called a man to discuss a persistent problem with his license plate.
The city’s AI-powered license plate readers, made by the tech startup Flock Safety, had repeatedly misread the man’s plate. At least six times, Flock had sent alerts to the Roseville Police Department, identifying his vehicle as stolen or used in a felony crime. Roseville police could see that Flock’s software kept confusing the “9” on the man’s license plate for an “8.”
The car owner said he would take off his license plate cover. Soon after, it happened again.
It’s “easier at this point we have it memorized,” a dispatch supervisor in Roseville wrote in an email to a colleague.
Andri Tambunan for BI
Flock says that in optimal conditions, its cameras accurately read more than 96% of license plate characters. Hundreds of pages of records from the Roseville Police Department show a different picture. In 2023 and 2024, Flock sent 1,427 alerts to Roseville police, flagging vehicles as stolen or used in a felony after they passed one of the city’s Flock cameras. An analysis by the police department found that in 71% of those alerts, Flock’s machine-learning software incorrectly read the license plates.
Business Insider filed public-records requests to obtain email correspondence, data, meeting notes, contracts, and other records that reveal the extensive back-and-forth between Roseville and Flock over repeated errors by Flock’s system. The cameras regularly missed vehicles, captured blurry images, misread license plate characters and states, and sent delayed alerts to police about vehicles possibly connected to crimes, the records show.
Roseville — which, Flock noted, has an atypical setup for its cameras— has flagged problems to the startup dozens of times over the past four years. The department said it has reported issues to the company as recently as this year, and it continues to meet with Flock twice a month to provide updates and share concerns.
Flock’s cameras, typically mounted on poles and street signs, use machine learning to capture license plate numbers and letters, as well as vehicle characteristics like color, model, dents, and bumper stickers. The Atlanta-based startup, which was recently valued at $8.4 billion, says it has cameras in more than 6,000 communities across the country, and it touts more than 20 billion vehicle reads each month.
Law enforcement agencies say Flock’s technology can help them track down vehicles and locate crime suspects. In some cases, officers can search for a specific car across tens of thousands of cameras nationwide.
The technology has also sparked a growing backlash. Dozens of communities have canceled, paused, or suspended contracts, citing privacy concerns, potential misuse of the cameras, and data sharing with federal immigration officials.
A factor that contributed to the misread problems in Roseville was the “particularly unique deployment” that the city requested, a Flock spokeswoman said. Roseville said it has its cameras configured so that they capture only the backs of vehicles, a setup intended to avoid capturing personally identifiable information like faces. Roseville’s setup included older hardware and placement of cameras higher and further from vehicles than the company typically recommends, Flock added.
The company told Business Insider that it is “dedicated to going above and beyond” to support customers in keeping their community safe, “even when the deployment is more challenging than our typical conditions.”
Max Isaacs, director of technology law and policy with the Policing Project at New York University’s School of Law, described Roseville’s misread data as “stunning.”
“A 71% misread rate is really beyond the pale,” he told Business Insider, referring to Roseville’s reported error rate for stolen and felony vehicle alerts. He added that “in the absence of concrete rules that directly address the problem of misreads, this technology is not safe to be deployed in communities.”
Andri Tambunan for BI
The consequences of a Flock camera misread can be devastating. As Business Insider previously reported, in at least a dozen instances in other jurisdictions across the country, misreads, or a lack of verification by officers, resulted in innocent people being stopped at gunpoint or detained in jail, among other outcomes. In Toledo, Ohio, driver Brandon Upchurch was mauled by a police dog after a Flock camera misread the “7” on his license plate as a “2.” As a result of his injuries, he said, he lost his job and was evicted from his home. He settled a lawsuit against the city and police officer for $35,000.
None of the incorrect Flock alerts in Roseville resulted in a traffic stop or arrest, a department spokesman said. Roseville requires police officers to verify that a license plate is stolen or have independent reasonable suspicion of a crime before making a traffic stop.
States including Montana, Virginia, Washington, and Kentucky have passed laws that require officers who receive alerts from a license-plate reader to independently verify the plate before initiating a stop, according to NYU’s Policing Project.
California doesn’t have such a requirement.
California Assembly Member Josh Lowenthal, who previously introduced legislation that would have put restrictions on license-plate reader data retention, called the Flock errors in Roseville disturbing and said they require immediate legislative action. He said he would call for a state-wide review of the reliability of license plate readers.
“If the data is flawed, then so too will be our justice system,” Lowenthal told Business Insider. License-plate readers, he said, “should make Californians safer, not put them at risk.”
Flock said Roseville’s camera performance has significantly improved, which the police department disputed. Roseville said that its data on Flock alert accuracy isn’t complete for 2025 or 2026, and that it hadn’t analyzed its recent data “to verify to what degree performance has or has not improved.”
The company declined to say whether it independently tests the accuracy of its cameras.
Andri Tambunan for BI
Roseville, which began using Flock cameras in 2021, said it continues to contract with the company because the cameras have helped the city solve crimes, including some cases that otherwise might have remained unsolved. The city, which last fiscal year had a police department budget of $69 million, has spent around $450,000 since 2020 on Flock products.
“Our agency acknowledges that technology is not perfect and requires human oversight,” the Roseville spokesman said. “We take that responsibility seriously.”
“We will continue to be steadfast in reporting issues as they arise, and committed to partnering with Flock to improve,” he added.
Frustration in Roseville
Flock’s cameras also missed vehicles altogether.
Last fall, Roseville police were trying to track down a person suspected of breaking into mailboxes the previous night. The suspect’s vehicle, a dark gray SUV, drove past a Flock camera, which missed it, Roseville records show.
The same thing happened in early January, when police were looking for a suspect in a deli break-in. And again, the following day, when a gray sedan was speeding and swerving recklessly on the road.
The frequent errors and misses by Flock’s cameras have led to frustration at the police department. Amanda Jones, a supervisor for the department’s real-time crime center, sent emails documenting vehicles and license plates that had been missed or misread by the cameras.
For example, in April 2025, Jones notified Flock employees that 16 of the 17 license plates that Flock had flagged the previous week were incorrect. They included what she described as “blatant misreads,” where the cameras mistook characters that weren’t similar, like reading a “3” as a “2.”
“Ouch!” replied a lieutenant, who was copied on the email.
In emails to the department, Flock apologized for the errors and promised improvements. Flock employees shared possible solutions, including reframing the cameras and introducing software updates to address issues like blurry plates and overexposed images.
At one point, Flock’s product director for machine learning suggested that officers ask drivers to remove license plate frames that made it “very difficult for the machine vision to tell it is actually a ‘E’ and not an ‘F.'” Roseville told Flock at the time that it wouldn’t do so, according to the department spokesman.
Flock in 2024 shared an analysis with Roseville’s police department, outlining reasons for the misreads. Vehicles far from a camera were sometimes blurry. Plates were cut off by trees or license plate frames. And Flock’s software confused similar characters, mistaking an “N” for a “V”, or a “1” for a “4.”
Roseville isn’t the first organization to flag inaccuracies in Flock’s technology.
In 2021, the research firm IPVM independently tested Flock’s license plate readers, concluding that Flock misidentified the state in about one of 10 reads, and that the system regularly misclassified vehicles’ type and make. IPVM said that Flock subsequently blocked it from purchasing its cameras for testing.
U.S. Sen. Alex Padilla of California told Business Insider that “inaccuracies stemming from law enforcement’s use of this technology are deeply alarming, as are the mass surveillance capabilities that it enables.”
Technologies like Flock, he said, “should only be deployed with strong safeguards that protect both public safety and Americans’ constitutional rights.”
Design, development, and graphics by Monsicha Srisuantang, Hongqian Li, and Randy Yeip.
Have a tip or a question about our reporting? Reach out to Business Insider’s enterprise team at investigations@insider.com.
