A collaboration between Lewis McLain & AI
The Fall the Cameras Went Dark
In about eight weeks, the license plate reader went from an invisible line item in the police budget to a campaign issue. In the last week of August, Governor Abbott ordered state agencies to stop funding Flock Safety cameras. Within days, Plano’s program ended, Princeton parked six cameras it had just received, and Wylie voted unanimously to end its contract.
Dallas turned off 321 state-funded cameras but kept roughly 300 city-funded ones; only three council members voted to cut the remaining $1.5 million. Pflugerville terminated after learning that 459 outside agencies had searched its data when the city had authorized 83. Bastrop ordered its cameras covered. League City and Conroe are talking about letting voters decide at the ballot box.
Nationally the story is the same, only louder. One advocacy group counts more than 200 cities and counties that have ended Flock contracts; a more conservative newspaper count found 82 terminations in 28 states through May. The Senate Judiciary crime subcommittee held a hearing in September titled “Always Watching.” Connecticut and Washington passed new limits. Flock cut its default data retention from 30 days to seven, made audits mandatory, laid off about 18% of its staff, and raised its federal lobbying spending to $360,000 for the quarter.
Much of the anger is earned, and I will get to why. But a city council that votes to unplug ought first to remember what it bought, and why. The original reason for these cameras was narrow, practical and good. That reason has not gone away just because the network built around it got out of hand.
The Original Reason: A Car Was Used in a Crime
Flock began in 2017 in an Atlanta subdivision, not a police headquarters. Garrett Langley, an electrical engineer, watched a string of break-ins and a stolen gun go unsolved in his own neighborhood. When he asked police chiefs and homeowner association boards why, the answer was the same: the bottleneck was evidence, and the most useful piece of evidence was a license plate.
The first customers were HOAs. One 37-home Minnesota development split a $2,000 annual subscription, about $54 per household, after a resident found a stranger in his kitchen. The cameras were solar powered, so nobody had to trench power or fiber to a neighborhood entrance. The pitch was modest: if a car comes in to commit a crime, we will have its tag.
Texas adopted the cameras for an equally specific reason. In 2023, with catalytic converter thefts spiking, the Legislature unanimously added $1 to every auto insurance policy to fight that crime. The bill was named for Harris County Deputy Darren Almendarez, killed after he interrupted converter thieves in a grocery store parking lot. When the Motor Vehicle Crime Prevention Authority offered grants, local agencies chose plate readers over investigator overtime and training by a wide margin.
The logic was sound. A converter theft is a quick, mobile crime with few witnesses, and the getaway vehicle is often the only lead. North Texas cities described the same narrow purpose when they signed up. Frisco told residents the cameras would alert officers to stolen vehicles, wanted suspects, and AMBER or Silver Alert vehicles, would not be used for traffic enforcement, and would not capture faces.
That is the baseline worth remembering. The camera was built to answer one question: was this car, tied to this crime, on this road?
What the Cameras Actually Do Well
Strip away the marketing and the panic, and the benefits fall into four plain categories.
First, they turn no-witness crimes into solvable ones. Car theft, converter theft, porch-pirate runs and smash-and-grabs rarely leave a witness who caught a tag. A camera at the neighborhood entrance or a major intersection supplies the one fact a detective needs to start. Texas’s vehicle crime authority reported that plate reader data was searched about 62,000 times in 2025 and led to roughly 1,660 cleared converter theft cases.
Second, they find people fast when minutes matter. Recent examples are not hard to find:
- Moody, Alabama: a 13-year-old taken from a convenience store was located on an interstate, unharmed, before an AMBER Alert could even be issued.
- Lucas County, Ohio: a deputy found a missing girl in under 30 minutes by running the suspect’s plate and seeing where it kept appearing.
- Toledo, Ohio: in October, a camera hit led police to one missing teen and then to a pregnant 16-year-old who had been missing since March.
- Greenville County, South Carolina: a woman and her son held by an armed boyfriend were tracked across North Carolina to Bristol, Tennessee, where he was arrested.
- Flagstaff, Arizona: cameras helped locate a missing toddler from California, months before that city canceled its contract.
Third, they multiply small departments. Prosper’s chief called his roughly 50 cameras a force multiplier. Even in Ingleside, where the council voted 6-0 to cancel, the discussion included praise for the system as a great tool. A town with a dozen officers cannot stake out every entrance; a camera can.
Fourth, they are cheap relative to the alternative. Frisco approved $434,150 for 81 cameras, about $5,360 per camera under that agreement. At that price, a dozen cameras come to about $64,300. The honest test is to set that figure beside the fully loaded cost of one patrol position (salary, benefits, vehicle, equipment) in your own city’s budget. The cameras also work nights and weekends.
The Numbers, Rebuilt Honestly
The defenders of these cameras do themselves no favors with headline claims. A fair case rests on the evidence that survives scrutiny, and that evidence points to a real but moderate benefit concentrated in vehicle crime and locating people.
| Claim | Who says it | What it actually shows | Weight |
|---|---|---|---|
| Crime reductions of up to 70% | Frisco PD release, repeating Flock marketing | A best-case ceiling with no stated base or comparison group | Weak |
| Involved in clearing over a million crimes last year | Flock’s CEO, on a podcast | “Involved” is undefined; no denominator | Weak |
| 9.1% higher clearance rate per camera per sworn officer | Flock-funded study with TCU and UT Tyler researchers, 123 agencies, 2023 | A regression association, paid for by the vendor | Weak to moderate |
| 11% fewer car thefts, 16% higher car-theft clearance | University of South Carolina working paper, 216 agencies | The first independent look at fixed networks; other property crime did not fall the same way | Moderate |
| Same study, caveat | The authors and the Institute for Justice | Thefts and clearances were already improving before the cameras went up, so not all of the gain belongs to them | Discount applied |
| Fewer shootings, car thefts and property crimes; no drop in violent crime | Atlantic City evaluation, Shjarback and Sarkos | A peer-reviewed single-city result | Moderate |
| 62,000 searches, about 1,660 cleared converter cases in 2025 | Texas Motor Vehicle Crime Prevention Authority | About 1 clearance per 37 searches; “led to” may overstate the camera’s role | Moderate |
| 13,000 searches a week on about 2,100 unique plates | Dallas Police, to council | About 6.2 searches per plate: targeted, repeat lookups on specific vehicles, not idle browsing | Useful context |
Read the table from the bottom up and a sensible picture emerges. The cameras do not cut crime 70%. They appear to cut car theft by something on the order of a tenth, lift vehicle-crime clearances, and give investigators a fast lead in abductions and pursuits. For many cities, at a few thousand dollars a camera, that is a good buy. It is not a miracle, and councils should stop being sold one.
Plano’s chief made the more defensible version of the argument this fall: the program was useful and crime fell over the three years it ran. That is correlation, not proof, and it is exactly why cities should collect their own outcome data rather than lean on the vendor’s.
What Went Wrong Was Drift, Not the Camera
Nearly every scandal driving the backlash traces to how the system was governed, not to the idea of reading a plate on a public street. Four kinds of drift stand out.
Sharing drift. A city buys cameras to watch its own entrances and quietly joins a national lookup network. Pflugerville authorized 83 agencies and found 459 had searched its data. A Shaker Heights, Ohio police chief admitted his department could only take the vendor at its word about who else was looking.
Purpose drift. Records obtained by 404 Media showed more than 4,000 searches between June 2024 and May 2025 listing ICE or immigration as the reason, run by local and state police on federal agents’ behalf, including in Illinois, where state law forbids it. Other logs showed more than 1,200 searches for suspected loitering. In Texas, police searched roughly 83,000 cameras, some in other states, for a woman suspected of a self-managed abortion. None of that is what a city council voted for when it approved cameras to catch converter thieves.
People drift. The Washington Post counted at least 50 officers charged with or accused of misusing plate readers, many to track wives, ex-girlfriends or women they wanted to meet. In Texas, a Lufkin officer faces 100 counts for allegedly tracking 11 people for more than a year. An Indianapolis officer allegedly searched his ex-wife’s plate more than 1,000 times, and that department had no routine audit of searches until recently.
Money drift. The Texas Tribune found the $1 converter-theft fee helped pay for at least 3,200 Flock cameras, many on highways and toll roads, and lawmakers said they never discussed that use when they passed the bill. When the governor stopped the money, cities like Plano and Dallas learned the program had never been fully priced into their own budgets.
Add accuracy problems, such as Roseville, California logging more than 1,000 false positives over two years, and the pattern is clear. A camera designed to answer whether this car was on this road became a searchable, 30-day memory of everyone’s car, open to thousands of users, for loosely stated reasons, with almost nobody checking. That is a governance failure. Governance failures have governance fixes.
Keeping the Benefit: A Ten-Point Council Ordinance
A council does not have to choose between a dragnet and darkness. It can adopt rules that keep the cameras tied to their original reason. Most of these already exist somewhere: in Connecticut’s and Washington’s new laws, in Michigan’s pending bill, and in the guardrails Flock itself announced in August.
- Purpose limit. List the allowed uses in the ordinance: stolen vehicles, felonies, missing or endangered persons, AMBER and Silver Alerts. Everything else is prohibited, including traffic enforcement and loitering.
- Short memory. Delete routine data after 7 days, Flock’s own new default, and never later than 21 days, the Connecticut and Washington standard. Keep longer only records tied to an open case.
- Sharing by name. No national lookup. The council approves, by name and once a year, each outside agency allowed to search, and requires the stated limits in writing, as Connecticut now does.
- Case number on every search. No case or incident number, no query. Flock now requires this; the city should require it too.
- Audit with teeth. A supervisor reviews searches monthly, an outside auditor reviews them yearly, and misuse is a firing offense. Langley has urged states to make illegal use of plate data a crime; Texas legislators should take him up on it.
- Warrant for tracking a person over time. A real-time hit on a stolen car needs no warrant. Reconstructing someone’s movements over days should, with exceptions for active emergencies and missing children, as Michigan’s bill proposes. Congressman Keith Self of McKinney has proposed a warrant rule for federal access.
- An annual public report. Cameras, searches, unique plates, outside agencies with access, hits, arrests and clearances by offense, false hits, and misuse cases. If the program works, the numbers will show it.
- Signs. Tell drivers when they enter a camera zone.
- Budget honesty. Carry the full cost as its own general fund line, voted in the open, so the program does not disappear the day a grant does. Add a sunset: the program expires every two years unless reauthorized with the report in hand.
- A clean contract. The city owns the data, the vendor gets no secondary use, and termination comes with a certified deletion.
A program run this way is very close to what residents thought they were getting in the first place.
Questions This Debate Has Not Answered
Several questions came up in the reporting that nobody has yet answered with data. A council weighing its vote, in either direction, should ask for each of them.
- What does unplugging cost? No city that canceled has published before-and-after numbers on vehicle theft, clearances or missing-person recoveries. Until someone does, both sides are arguing from anecdotes.
- Does canceling a city contract actually remove the network? Often not. The Department of Public Safety installed nearly 1,200 cameras of its own under a $15.9 million contract, and Austin police reportedly kept using that state network after the city halted its program. In Princeton, most of the Flock cameras are privately owned by neighborhoods, and Fort Worth lets residents register private cameras for police use.
- What happens to the data after a city cancels? South Lyon, Michigan ordered its police to stop accessing and sharing the data. Few cities have said publicly whether the vendor certified that the data was deleted.
- Does switching vendors solve anything? Sierra Vista, Arizona canceled Flock and is shopping for another plate reader vendor with the same grant. Witnesses at the Senate hearing urged that any rules apply to Axon and Motorola as well. Princeton already runs Motorola Vigilant cameras. The governance problem follows the data, not the brand.
- Who pays now? The governor’s office said most local camera money is federal. The Texas Tribune documented at least $45.9 million in state money, counting the vehicle authority grants and the DPS contract. Nobody has published the federal share, and cities that keep their cameras now carry the full cost locally.
- Is seven days a rule or a suggestion? Coverage of Flock’s August changes conflicts. Some reports call seven days the new default; others say Flock is recommending it while agencies keep the authority to set their own. A council should write its own number into ordinance rather than rely on a vendor setting.
- How accurate are the hits? Roseville’s false positives surfaced only through an internal review. Almost no city reports its false-hit rate, even though a bad hit can mean a felony stop on an innocent driver.
- What will Texas law say? City-by-city rules are a patchwork, and the debate has become a statewide campaign issue. The Legislature returns in January 2027, and the converter-fee statute that paid for so many of these cameras is a natural place to start.
What the Critics Would Say Back
A fair essay owes the other side its strongest arguments, and they are serious.
The constitutional objection. The Institute for Justice, the ACLU and EPIC argue that a searchable record of every driver’s movements, queried without a judge’s approval, is a warrantless dragnet. On that view, shorter retention and audits make the dragnet smaller but do not make it lawful. Several pending lawsuits, including one in San Diego, rest on this argument.
The trust objection. Local rules only work if the vendor and the network honor them. Misuse has mostly been uncovered by reporters, public records requests and citizen tools, not by internal audits. A Pinal County, Arizona sheriff told the Senate the company first said its cameras used no AI and later contradicted that. Connecticut found it had to change state law and the vendor’s software to stop out-of-state access.
The evidence objection. The best independent study is one working paper with a pre-trend problem, and most of the rest is vendor-funded or anecdotal. Critics say councils are being asked to accept a permanent surveillance cost for a benefit that has not been proven.
The consent objection. Polling cited by The Texan found nearly 64% of Texas likely voters oppose plate cameras at public intersections, and a national poll found unfavorable views of about 40% among Democrats and Republicans alike. San Diego Councilmember Stephen Whitburn put it this way, in substance: a safety tool that makes residents feel less safe has stopped doing its job.
Those arguments explain why so many councils have chosen to unplug rather than reform. The question for every council is whether rules like the ten above can earn back that trust, or whether the trust is already gone.
Sources
- Texas Tribune via KXAN: Abbott blocks state agencies from spending on Flock cameras
- Texas Tribune via ABC13: $1 insurance fee paid for thousands of Flock cameras
- Straight Arrow News: state fund totals, including the DPS contract
- San Antonio Current: converter-fee grants and 2025 clearance figures
- Community Impact: Plano’s Flock contract ends
- FOX 4: Princeton won’t use Flock cameras; Wylie ends contract
- Princeton Herald: most current Flock cameras in Princeton are private
- FOX 4: Dallas council rejects ending Flock funding
- The Texan: Texas cities and counties break with Flock
- CBS Austin: Bastrop covers its cameras
- Newsweek: League City and Conroe weigh a public vote
- KRLD: Flock cameras across DFW and Fort Worth’s community camera program
- Community Impact: Frisco’s 81-camera agreement
- City of Frisco: Flock camera announcement
- Killeen Daily Herald: Ingleside cancels Flock contract
- CBS Minnesota: HOA adopts Flock; founder’s story
- Flock Safety media kit: founding story
- ABC 33/40: Moody police find abducted 13-year-old
- WTVG via Cleveland 19: Toledo recovers three missing children
- WTVG via WAGM: Lucas County deputy finds missing girl
- Post and Courier: Greenville case and USC study
- City Journal: USC authors on their study
- Institute for Justice: critique of the USC study
- Shjarback and Sarkos: Atlantic City ALPR evaluation
- Police1 via STIRM: Flock-funded clearance study
- Washington Post via Anchorage Daily News: officers misusing plate readers
- Tech Brew: abortion search and other cases
- Reason via Yahoo: 404 Media immigration search records
- Cybernews: loitering searches
- Cleveland Scene: Shaker Heights and network sharing
- Michigan Advance: South Lyon covers cameras and ends contract
- KGUN 9: Sierra Vista ends contract, seeks another vendor
- Northeastern University: 200+ terminations, retention cut to 7 days
- GovTech: 82 terminations in 28 states
- FOX 6: Flock’s mandatory audits and new retention
- Straight Arrow News: Flock’s new guardrails and agency authority over retention
- Live 5 News: Senate subcommittee hearing
- Fox Carolina: hearing testimony on Axon and other vendors
- Bloomberg Government: Flock lobbying spending
- ValueAdd VC: Flock layoffs
- TechCrunch: Langley calls for compromise
- WFSB: Connecticut’s new plate reader rules
- Washington State Standard: Washington’s new law
- WNEM via WNDU: Michigan warrant bill
- Times of San Diego: council resolution and ACLU suit
- Roll Call: hearing preview and polling
- TechSpot: Roseville false positives
- Slashdot / Washington Post: Indianapolis officers charged