Flock Exposed - Claims vs Reality

Flock markets itself heavily on its recovery and success numbers. Before your Town Board approves a contract based on these figures, it's worth looking at what stands behind them.

The "10% of crime solved" claim - and the study's own academic co-author now has doubts

Flock states its technology helps resolve roughly 700,000 crimes a year, or about 10 percent of reported U.S. crime, a figure also cited in some Flock materials as up to 20 percent of cases agencies solve. What Flock does not clearly explain is how that percentage is calculated, or what counts as a Flock camera "helping" solve a case versus simply being one small piece among many pieces of evidence. A fact-checking review by DeFlockILM of Flock's own "Myths & Facts" marketing material found that this figure and others like it are Flock's own marketing numbers, not independent findings, and that the "solved" claim rests on a quiet substitution: a case where a camera was simply queried gets counted the same as a case the camera actually helped solve. The study itself was conducted by Flock's own employees and given an appearance of outside legitimacy through the "oversight" of two named academic researchers - but one of them, Johnny Nhan of Texas Christian University, later told 404 Media that "I personally would have done things much differently" than Flock's researchers did, and said the police data underlying the study was "too varied and incomplete" for meaningful statistical analysis, leading him to change direction on his own future Flock research. Cleveland's own Cuyahoga County Prosecutor, an outspoken Flock supporter, illustrated the same problem from the inside: he told City Council that producing a real count of Flock-assisted cases would require manually reviewing thousands of investigative files just to determine whether a camera generated the original lead, merely corroborated other evidence, or played only a minor role - meaning even Flock's strongest backers cannot cleanly say what actually counts as a "solve."

Source: DeflockILM, "Flock Safety's 'Myths & Facts,' Fact-Checked," 2026; Government Technology, "How Many Crimes Has Flock Safety Solved in Cleveland?," 2026; 404 Media, "Researcher Who Oversaw Flock Surveillance Study Now Has Concerns About It"

A "70% crime drop" that didn't hold up to a wider look

San Marino, California is a case Flock itself publicized. For a brief period in 2021, the affluent Los Angeles County town saw a sharp drop in residential burglaries, its most common crime, and Flock seized on the numbers, claiming its cameras were key to cutting overall crime in the town by 70 percent. A Forbes investigation that took a wider view of the same town's crime data found a very different story: despite the initial dip, crime in San Marino went up, not down, undercutting the very statistic Flock had circulated to sell the system to other communities.

Source: Forbes, "Flock Installed AI Cameras In This Small City And Claimed Crime Went Down. It Went Up.," 2024

A Dayton success story that predates the cameras themselves

Flock frequently touts Dayton, Ohio's Twin Towers neighborhood as a success story, citing a 43 percent decline in crime the company attributes to its cameras. A Forbes investigation traced that statistic back to a Dayton Police Department report and found the 43 percent drop was measured across the March 1 to July 20 window comparing 2018 to 2020 - but Flock's cameras were not installed in Dayton until March of 2020. The crime decline Flock credits to its own technology had already happened before the technology was in place.

Source: Forbes, "Flock Installed AI Cameras In This Small City And Claimed Crime Went Down. It Went Up.," 2024

Atlanta added 5,000 cameras. The city's own FBI-reported numbers show clearance rates falling, not rising

Atlanta has more Flock and other surveillance cameras per resident than almost anywhere else in the country - more than 5,000 Flock units alone, feeding into a citywide network of roughly 28,000 cameras overall. If the technology delivered on its central promise, the city's crime clearance rates should show it. Instead, an Atlanta Community Press Collective analysis of FBI crime data found that Atlanta Police Department's clearance rates for homicide, robbery, miscellaneous larceny, and shoplifting all declined between 2021 and 2025, the same period the camera network expanded roughly eight-fold. The homicide clearance rate alone fell from 53.4 percent in 2021 to 48 percent in 2025. This is the same city where Flock's own CEO publicly touted an Atlanta detective who said Flock helped solve 35 homicides in a single year - a claim that sits uneasily next to the city's own declining clearance rate over that same broader period.

Source: Atlanta Community Press Collective, "28,000 cameras and counting, but Atlanta's clearance rates aren't rising," July 2026; Futurism, "Flock Surveillance Cameras Are Actually Horrible for Fighting Crime, FBI Data Shows," 2026; The Cool Down, "Atlanta added 5,000 plate readers, and FBI data shows crime-solving barely budged," 2026

Cleveland's own police department couldn't tell the city council how many crimes Flock actually solved

Ahead of a Flock contract renewal vote, the Cleveland Division of Police and the Cuyahoga County Prosecutor's Office were unable to tell City Council how many cases had actually been solved with the technology's help. In a July 2026 memo, police told council that those numbers simply are not tracked, writing that police reports and case materials "are not written or maintained for the purpose of allowing a researcher to later isolate and measure the efficacy of a single investigative tool," since investigations typically combine license plate data with surveillance video, witness statements, forensic evidence, and other evidence rather than relying on any one tool. Council members, who had asked for hard numbers rather than success anecdotes, voted to extend the contract for only six months instead of a full renewal, citing continued discomfort with the lack of verifiable data.

Source: Government Technology, "How Many Crimes Has Flock Safety Solved in Cleveland?," 2026; Signal Cleveland, "Cleveland City Council poised to back six-month extension for Flock license plate readers," 2026

A 2026 "11% drop in car theft" study left out the one number that didn't help Flock's case

Flock publicized a 2026 independent academic study claiming vehicle thefts fell 11 percent after its cameras went live, calling it "a model for the kind of research we want to see more of" and noting the study used national FBI crime data. But reporting on the same study found that Flock's own summary left out a third outcome the researchers had actually measured: how quickly stolen vehicles were recovered. Looking at the full population of stolen vehicles, including ones never recovered at all, the study's authors found no statistically detectable improvement in recovery speed at the one-day, seven-day, or thirty-day marks. Flock's blog post also cited its own 2026 Customer Census, a self-reported survey of its own client agencies, as corroborating evidence - which is not independent corroboration, but the company citing its own marketing research to support a study about its own product. The same academic paper separately criticized this genre of evidence, describing vendor-supplied comparisons of adopters and non-adopters, including a prior Flock-published analysis, as failing to address why some agencies adopt the technology and others do not in the first place.

Source: Yahoo News, "Flock Safety Claims an 11% Drop in Car Theft. Here's What It Left Out"; Flock Safety, "Vehicle Thefts Fell 11% After Flock Cameras Went Live"

Accuracy numbers that don't add up

Flock's accuracy figures are self-reported, and the company has not published a methodology, a sample size, or test conditions beyond the phrase "optimal deployment conditions." Separately, in litigation, Flock has stated its cameras accurately capture 93 of every 100 plates, a figure that conflicts with other accuracy numbers the company has published elsewhere, with no explanation offered for the difference. A watchdog organization that independently tests security and surveillance technology, known as IPVM, has stated publicly that it repeatedly offered to pay full price for Flock hardware in order to test it independently and was blocked from purchasing it; Flock disputed IPVM's findings and halted sales to IPVM altogether. Roseville, California's own police department found that 71% of its Flock stolen vehicle and felony alerts, 1,427 of them over two years, contained a misread license plate.

Source: Red Banyan, "Flock's 71% Problem: What the Accuracy Numbers Really Mean"; TechTimes, "Flock Safety Crosses 100,000 Cameras as 53 Cities Cancel Over Unauthorized Federal Data Access"

Independent audits tell a different story

When outside researchers have actually been able to review Flock's real-world performance, the results have not matched the company's marketing. An LAPD Inspector General audit found that 161 of 498 hotlist alerts reviewed were false, a 32.3 percent error rate, and the department declined to renew its Flock contract as a result. At least 82 municipalities across 28 states have terminated their Flock Safety contracts citing privacy, accuracy, and immigration-enforcement concerns. Flock also states that only a small fraction of the vehicle alerts it sends to law enforcement are flagged as inaccurate, a claim that the trade publication Security Management said it was unable to independently verify.

Source: TechTimes, "LAPD Cuts Off Flock Safety After Audit Finds One-in-Three Hot-List Alerts Were False"; SAN, "LAPD cancels Flock contract after cameras falsely flagged vehicles as stolen"; Gadget Review, "One in Three Alerts Wrong: High Error Rates and Data Concerns Force Cities to Drop Flock"; ASIS International, Security Management, "Clocking Flock: Amid Public Backlash, Flock Safety Releases Accountability Features Dependent on Customer Enforcement"

A city's own security review of Flock was quietly altered before a contract renewal vote

Dunwoody, Georgia offers a documented example of the numbers themselves being manipulated, not just left unverified. Ahead of an April 2026 city council vote to expand its Flock contract by roughly 215,000 dollars, the Dunwoody City Council asked a city official to conduct an independent security assessment of Flock, specifically to reassure a public already unsettled by earlier controversies. Internal emails later obtained through open-records requests showed that the assessment was not independent at all: the city's Technology Director asked Flock's own salesperson to help "bring the score up," the scoring rubric was rewritten before the public presentation, a new favorable scoring tier was inserted, definitions were softened, and language recommending "full transparency to potential customers" was stripped from the evaluator's findings before release. The scores that reached the council and the public were drawn from answers Flock provided about itself, and no independent party ever verified them. The city council voted unanimously to approve the contract expansion anyway. The manipulation only came to light because a local resident, Jason Hunyar, filed the open-records requests that surfaced the internal emails.

Source: IPVM, "City and Flock Manipulate Security Scores," 2026; Government Technology, "Why Flock Safety Finds Itself in a Surveillance Backlash," 2026; Jason Hunyar, "Flock's Secret War Council," FOIA Around And Find Out, 2026

That same Dunwoody records fight also turned up something separate from the doctored score

Flock's own sales staff had been logging into live camera feeds at a local Jewish community center - including a children's gymnastics room and a swimming pool - to demo the system's capabilities to prospective police customers, months before the contract vote in question. Flock's CEO later apologized for what he called "poor judgment," but the council had already renewed the contract without knowing that had happened.

Source: 404 Media, "City Learns Flock Accessed Cameras in Children's Gymnastics Room as a Sales Pitch Demo, Renews Contract Anyway," 2026; Gadget Review, "Flock Safety Employees Watched Kids' Gymnastics Room to Pitch Surveillance Tech," 2026

Flock will not let outsiders check its own numbers

It is not just that Flock's statistics lack a published methodology. In multiple documented cases, the company and its law-enforcement customers have actively kept independent parties from checking the claims at all. IPVM's inability to simply buy Flock hardware at full price for independent testing is one example. In Palo Alto, California, the city's own regular auditor, Baker Tilly, had to recuse itself from a planned audit of the Flock camera system because of a conflict of interest arising from Baker Tilly's own work with Flock, delaying the review and forcing the city to bring in a separate independent police auditor, OIR Group, instead. Flock's own August 2026 privacy overhaul made its "Audit Assistance" self-monitoring tool mandatory for all law enforcement customers, but the ACLU has publicly stated that no independent evaluator has verified the tool actually works, meaning the public has only Flock's word that it catches misuse - and a separate ACLU of Massachusetts investigation of Flock network audits found police frequently enter vague terms like "investigation" or "susp" instead of any substantive reason into the mandatory search-reason field, undercutting Flock's own claim that requiring a reason meaningfully deters misuse. Flock has also commissioned a security review of its own systems by an outside firm, Bishop Fox, but has said it will publish only a summary of that review's findings rather than the full report, so the public and researchers cannot examine the underlying data themselves.

Source: Red Banyan; Palo Alto Online, "Flock camera study delayed in Palo Alto after auditor recusal" and "Palo Alto set to begin Flock audit amid growing public backlash"; San José Spotlight, same headline; TechTimes, "Flock Safety Overhauls Rules, But Key Audit Fix Has No Independent Check"; ACLU, "Despite 'New' Updates, Flock's Creepy Cameras Remain Major Civil Liberties Threat"; ACLU, "Flock Safety Credibility Lost as it Repeatedly Lies to City Councils, Police Departments, and Public Across the Country"; Carscoops, "Flock's Big Privacy Overhaul Comes Down To One Word Doing A Lot Of Work"

The "self-policing" audit tool

Flock has promoted its "Audit Assistance" tool as meant to catch officers misusing the system for personal reasons. But as noted above, the ACLU has pointed out that while dozens of officers have been arrested, fired, or disciplined for misusing Flock, no independent evaluator has verified whether the audit tool is actually catching most violators, or only a small fraction of them. In other words, the tool has not been shown to work, only that Flock says it does. It's also worth noting that the police departments using the system are largely the ones auditing their own officers' use of it. In practice, this means Flock is auditing itself, and police departments are auditing themselves, the equivalent of the fox watching the hen house.

Source: ACLU, "Despite 'New' Updates, Flock's Creepy Cameras Remain Major Civil Liberties Threat"

The bottom line: Flock's numbers come from Flock, and when outside parties have tried to check those numbers independently, they have repeatedly been blocked, delayed, manipulated, or given only a summary instead of the full picture. Even a Flock-publicized study's own named academic reviewer has since distanced himself from how it was done. Before your town signs a contract based on these claims, ask who has independently verified them, and what happens to the data long after the "solved" case is closed. Flock's default retention period is now seven days, down from a previous thirty, but that is only a default. The company's new "Evidence Mode" feature preserves any search tied to a case number instead of letting it age out, and according to the ACLU, if that applies broadly to all data touched by a search rather than narrowly to specific evidence, it could functionally mean indefinite retention of ALPR data collected nationwide.

Source: ACLU, "Despite 'New' Updates, Flock's Creepy Cameras Remain Major Civil Liberties Threat"

This page will be updated as more factual sources become available.