08/17/2026
👇🏼
📸 THE CAMERAS WENT DARK 📷
Three license-plate readers in Hood County are now disconnected and covered. The constable who helped shut them down says the technology works. His argument is that America built the network before it built the rules.
August 17th, 2026AD
By Hood County Express
The cameras were still standing when they went dark.
One along East U.S. 377. Another on Tin Top Highway. A third near State Highway 144 and Mitchell Bend Highway.
They were not smashed, cut down or hauled away. The Hood County Precinct 2 Constable’s Office disconnected the three automated license-plate readers and placed non-damaging coverings over them, leaving the equipment where it stood until it can be physically removed.
The scene lends itself to an easy story: a Texas constable taking government surveillance into his own hands.
That is not what happened.
On August 11th, the Hood County Commissioners Court voted 4-0 to authorize the Precinct 2 Constable’s Office to disconnect and temporarily cover the county’s three remaining contracted Flock Safety camera units following termination of the county’s agreement. The coverings were to remain until the equipment was removed.
Constable John D. Shirley carried out the order.
Four months earlier, on April 14th, commissioners had voted unanimously to terminate the Flock arrangement and seek the return of approximately $126,000. By June, the dispute had escalated far enough that an official Commissioners Court agenda called for possible legal action against Flock over what the county described as the company’s refusal to issue a refund. The ultimate disposition of that money was not confirmed before publication. (Hood County Documents on Demand)
Shirley makes no effort to hide the fact that he wanted the cameras gone.
But he draws a hard line between wanting something as an elected official and possessing the authority to do it as a peace officer.
“I’m a police officer,” Shirley said. “I operate under the law.”
Had Commissioners Court voted to keep the cameras, he said, the answer would have been equally simple.
“No matter how badly I can’t stand them, I absolutely would have done what the law said.”
That concession is what makes the Hood County story more interesting than another fight over roadside cameras.
Shirley does not claim Flock has never caught a criminal. He does not argue that automated license-plate readers have no legitimate place in policing. He does not say every use of the technology violates the Constitution.
Ask him whether Flock has helped solve crimes and the answer comes quickly.
“Absolutely,” he said. “It would be foolish to say that no crimes have been solved.”
Hood County’s own experience supplies evidence for that side of the argument. The Sheriff’s Office has cited the technology in connection with the recapture of three escaped juvenile offenders, a homicide-related investigation and the recovery of an elderly woman reported missing from Fort Worth. (Hood County Today)
Shirley’s objection begins after that point.
His question is not whether the technology can work.
It is how much power government should acquire because it works.
A CAMERA IS NOT A NETWORK
A license plate has never been private in the ordinary sense.
Drivers display one because the law requires it. Police officers have always been able to sit beside a highway, read a plate, check it against government records and follow a vehicle they have lawful reason to investigate.
But a police officer beside a highway forgets.
A computer does not.
That difference — memory, multiplied by scale — is where the modern argument begins.
Automatic plate readers are not new. Researchers at George Mason University have documented their rapid adoption by American police agencies for years. What has changed is the scale, cost and connectivity of the systems around them. Modern ALPR platforms can turn a fleeting roadside observation into a searchable record containing a vehicle image, plate, time, location and physical characteristics. (Springer)
Flock has built the most visible version of that architecture. The company’s technology is now used in more than 6,000 communities and by more than 5,000 law-enforcement agencies across 49 states. The Washington Post has reported that Flock’s cameras scan roughly 20 billion license plates a month. (The Washington Post)
This is not one national government database. Flock says individual customers own and control their data, decide who receives access and choose whether to establish sharing relationships with other agencies. The company says there is no automatic nationwide sharing. (Flock Safety)
But interconnected agencies can create something that no officer with binoculars ever could: the ability to search observations made far beyond a single patrol district, long after the vehicle itself has disappeared.
Where was this truck yesterday?
Where else has it appeared?
Was it near the crime scene before investigators knew to look for it?
What other cameras saw it?
For detectives working an abduction, homicide, robbery or missing-person case, that capability can be extraordinary.
For everyone else, its existence raises another question.
What rules govern the search?
A camera sees.
A network remembers.
And once a network remembers, government can ask questions of the past that once would have required teams of detectives, days of surveillance and extraordinary luck.
Until this month, Flock’s standard retention period was 30 days unless a customer agreement provided otherwise. (Flock Safety)
Then the company changed course.
THE GUARDRAILS ARRIVED LATER
On August 13th — four days before this report — Flock announced the most substantial tightening of its privacy and accountability controls in the company’s history.
Flock said it will shorten its standard retention period from 30 days to seven. Law-enforcement users will be required to enter a case code tying searches to an investigation, with emergency overrides flagged for review. An abnormal-search auditing tool that had previously been optional will become mandatory, with suspicious activity capable of locking a user out pending review. Customers will also receive more control over what types of offenses outside agencies may use to search their data. Many of the changes are scheduled to become mandatory by January 1st. (AP News)
Those changes did not occur in a vacuum.
They followed a year of intensifying controversies involving unauthorized police searches, interstate data-sharing problems, litigation, public-record investigations and a widening political backlash against ALPR networks. More than 50 agencies or jurisdictions have canceled, suspended, rejected or deactivated Flock systems since the beginning of this year, according to a tracker cited by the Associated Press. (AP News)
Flock itself acknowledged that some California law-enforcement camera networks were inadvertently accessible to out-of-state agencies during part of 2025. The company said earlier limitations in its technical logging made it impossible in some cases to determine exactly how the sharing occurred. Flock said it accepted responsibility and subsequently strengthened sharing controls and logging. (Flock Safety)
That sequence is central to Shirley’s argument.
Deploy the technology.
Scale the network.
Discover where the system can fail or be abused.
Then strengthen the rules.
Shirley thinks government has the order backward.
“Put the brakes on,” he said.
“Build the safeguards in on the front end.”
Texas had an opportunity to consider something close to that model.
House Bill 961, filed during the 2025 legislative session, would have required a warrant or court order for law-enforcement use of an automatic plate reader. It would have restricted ALPR information to investigations of violent offenses, required prompt destruction of unrelated data, restricted sharing and required agencies to publicly report how many plates they scanned and how many scans contributed to an arrest or prosecution.
The bill never advanced beyond referral to committee. (Texas Legislature Online)
The cameras did.
Washington took a different path this year.
Its new law generally limits ALPR retention to 21 days, restricts data collection around sensitive locations including places of worship and schools, prohibits using ALPR systems to track First Amendment-protected activity and requires a probable-cause warrant for agencies seeking privately held ALPR data.
And it contains one provision written for the moment when a database result becomes a police encounter:
An ALPR match, standing alone, is not reasonable suspicion for a traffic stop. Officers must develop independent reasonable suspicion or visually verify the plate and confirm its status through another authorized source. (LawFiles)
That rule is easier to understand after meeting Brandon Upchurch.
WHEN A HIT BECOMES A GUNPOINT STOP
In April 2024, Upchurch was driving a Dodge Ram in Toledo, Ohio, when a Flock camera read a “7” on his license plate as a “2.”
The wrong character caused the truck to be associated with a stolen plate.
Police stopped Upchurch. During the encounter, an officer released a police dog, which bit him. Officers later realized the ALPR information did not match the actual plate. Upchurch was hospitalized and jailed; the charges were eventually dropped. Toledo police later reprimanded an officer for failing to visually verify the plate before acting on the reader information. (ABC News)
Business Insider examined police records, lawsuits and local reporting around the country and documented a dozen cases in which Flock misreads or failures by officers to verify ALPR information contributed to innocent people being stopped at gunpoint, jailed or otherwise subjected to serious police encounters. (Business Insider)
That does not mean every bad Flock-related stop begins with a bad camera read.
Some do not.
An ALPR can read a plate perfectly and still produce the wrong outcome if an underlying hotlist is stale, the wrong vehicle has been entered, or an officer treats an alert as a conclusion instead of a lead.
The failure chain can run through several hands:
camera, algorithm, database, alert, officer.
A mistake at any one point can arrive at the roadside carrying the apparent authority of all the others.
Flock says officers should manually verify ALPR hits before enforcement action. That principle matters because the machine does not make the ultimate decision to draw a gun, apply handcuffs or release a police dog.
A person does.
The lesson is not that computers make mistakes and humans do not.
Humans make plenty.
It is that computers can deliver machine and human mistakes faster, across greater distances and with an appearance of precision that can discourage skepticism at exactly the moment skepticism is needed.
THE PEOPLE WITH THE PASSWORD
Cybersecurity creates another category of risk.
Independent security researchers have found genuine vulnerabilities in Flock hardware and software. Federal National Vulnerability Database entries describe versions of Flock applications used on Falcon and Sparrow readers with unauthenticated administrative endpoints, hard-coded cryptographic material and development Wi-Fi credentials stored in production firmware. One documented flaw could permit shell access to an attacker already on the relevant local network. (NVD)
Those findings should not be exaggerated.
They do not establish that someone sitting at a laptop anywhere in the world can casually break into Flock’s national infrastructure.
Flock says the hardware vulnerabilities disclosed by researchers required physical access and detailed knowledge of the devices, did not expose its cloud environment and did not compromise customer data. The company says its cloud platform has never suffered a breach in which customer data was accessed or exfiltrated by an attacker. (Flock Safety)
The more unsettling security problem may be the person who does not need to hack anything.
He already has a password.
A Washington Post investigation reviewed police and court records involving nearly 50 officers accused or charged with unauthorized ALPR use. Many of the cases involved officers using the technology to search for current or former romantic partners, relatives or other people for reasons unrelated to legitimate police work. (AP News)
The cases are not all alike. Allegations are not convictions. Some resulted in criminal charges, some in discipline or termination, and some remained under investigation.
But together they expose a weakness that cannot be solved by improving optical-character recognition.
The authorized user can become the threat.
Police officers have long possessed access to sensitive systems: criminal histories, driver records, warrants, investigative files. Misusing them is prohibited.
ALPR networks add something different.
They can answer a question about movement.
Where has this vehicle been?
That can turn an illicit database search into physical surveillance without the officer ever leaving a chair.
Flock argues that its audit trails are part of the answer. The company says auditing has identified misuse, including incidents that led agencies to investigate their own employees. Its new mandatory abnormal-search detection is designed to catch suspicious patterns sooner. (AP News)
That defense has merit.
An audit trail that catches misconduct is better than no audit trail.
But an audit that discovers months of improper tracking is not the same thing as a safeguard that prevents the tracking from happening.
And Flock’s decision to make some previously optional protections mandatory invites an unavoidable question:
If those protections are important enough to require now, should they have been required while the network was growing?
WHEN EVERYTHING WORKS
The deepest argument over mass surveillance does not begin when the machine fails.
It begins when everything works.
Suppose every plate is read correctly.
Suppose every database is current.
Suppose nobody hacks a camera.
Suppose every officer follows policy and nobody searches for an ex-wife, neighbor, political opponent or curiosity.
Government still possesses the capability.
That capability is the point.
Flock’s own 2025 Impact Census illustrates the case for it. Based on a survey of nearly 700 agencies along with company data and estimates, Flock says its technology supported roughly one million investigations or incidents in 2025 and assisted in an estimated 20 percent of cleared cases in responding customer jurisdictions. The company also says its systems helped locate more than 10,000 missing people. (Flock Safety)
Those are Flock’s estimates, not an independent national effectiveness study.
Independent research is more cautious.
A peer-reviewed review published by researchers including Cynthia Lum and Christopher Koper concluded that ALPRs are undeniably faster than humans at scanning plates and can assist with stolen-vehicle recovery and criminal investigations. But the researchers found only a limited number of rigorous studies measuring broader public-safety effects, with many producing mixed or modest results. They concluded that effectiveness depends heavily on how agencies deploy, supervise and integrate the technology. (Springer)
That does not mean Flock does not work.
It means “works” is not a complete policy analysis.
A wiretap works.
A search warrant works.
An undercover officer works.
A helicopter works.
Government still surrounds powerful investigative tools with rules because effectiveness is only one part of the bargain.
Shirley frames the question in plainer language.
How many crimes, he asks, are being solved with Flock that could not otherwise have been solved — or could not have been solved with reasonably comparable efficiency — using traditional investigative methods or less intrusive technology?
That is an empirical question.
He believes government should know the answer before asking citizens to accept the surveillance power in exchange.
WHAT THE CONSTITUTION HAS — AND HAS NOT — SAID
The easiest way to make the Flock debate less intelligent is to pretend the constitutional question has already been answered.
It has not.
The United States Supreme Court has never held that a Flock-style license-plate-reader network is unconstitutional.
A federal judge considering Norfolk, Virginia’s system ruled in January that the 176 cameras before the court did not capture enough of a person’s movements to amount to the kind of comprehensive tracking protected by Supreme Court location-privacy cases. The plaintiffs have appealed. (Justia Law)
The court’s reasoning matters because it identifies the unresolved line.
A single camera does not follow a person everywhere.
A scattered network may not either.
But what happens as the gaps disappear?
The Supreme Court has spent the last decade warning that digital scale can change the Fourth Amendment analysis.
In Carpenter v. United States, the Court held that government acquisition of historical cellphone-location information can constitute a search because it creates an intimate record of a person’s physical movements.
Then, on June 29th of this year, the Court decided Chatrie v. United States. Police investigating a bank robbery had obtained location information from Google identifying phones near the scene. The Supreme Court held that obtaining Chatrie’s cellphone location history was a Fourth Amendment search even though police sought only a limited period of information from a third-party technology company. The Court sent the case back for further consideration of whether the warrant itself satisfied the Fourth Amendment. (Supreme Court)
Flock is not a cellphone.
A roadside plate reader is not a Google geofence.
A car traveling on a public road is plainly visible to people standing nearby.
Those differences matter.
But Chatrie also reaffirmed something larger: location information can reveal far more than geography. Even short-term tracking, the Court noted, may expose details of familial, political, professional, religious and sexual associations. (Supreme Court)
That is where the plate-reader question becomes difficult.
One officer seeing a car outside a church tells government very little.
A database showing the same car there every Sunday may tell it more.
One observation of a vehicle outside a political meeting means almost nothing.
A searchable pattern can mean something else.
At what density, over what period, with what search capability, does the aggregation of lawful public observations become constitutionally different from seeing each observation separately?
No Supreme Court case has given America a definitive ALPR answer.
The technology did not wait for one.
THE RIGHT TO SHOW UP
There is another consequence of surveillance that is harder to count than arrests or stolen-car recoveries.
People sometimes behave differently when they believe they are being watched.
Law has long had a name for that phenomenon: a chilling effect.
Government does not always have to prohibit speech or association to burden it. Sometimes people retreat from lawful activity because they fear who may be watching, what may be recorded or how that information may later be used.
Plate readers did not invent that problem.
History shows why the location of a vehicle can matter.
An Associated Press investigation more than a decade ago revealed that New York police had photographed mosques, collected the license plates of worshippers and, in some instances, driven electronic plate readers past mosques to record vehicles parked nearby. Police records showed worshippers being monitored even when there was no identified criminal activity tied to the mosque. The AP investigation later won the Pulitzer Prize for investigative reporting. (The Associated Press)
That was not Flock.
Neither was another case that surfaced in 2016, when the chairman of the Senate Homeland Security and Governmental Affairs Committee demanded answers after reports that Immigration and Customs Enforcement had encouraged California law enforcement to record plates at gun shows in an effort to identify illegal fi****ms trafficking. The senator specifically asked whether the government had created a database of people attending those events. (Homeland Security Committee)
Again, not Flock.
The concern is older than the company.
The modern network simply makes the capability easier to scale.
And location can reveal association without proving it.
A car outside a church does not establish the registered owner was worshipping there.
A plate at a gun show does not establish its owner bought a firearm.
A vehicle outside a medical clinic does not establish why anyone entered the building.
But repeated location creates inference.
Church.
Mosque.
Political meeting.
Protest.
Addiction-treatment center.
Medical clinic.
Union hall.
Gun show.
Immigration lawyer.
The observation of a car at any one of those places may mean almost nothing.
A pattern can mean much more.
Washington’s new law recognizes that distinction by specifically restricting ALPR collection around places of worship and other sensitive locations and by prohibiting use of the systems to track constitutionally protected activity. (LawFiles)
There is no credible evidence reviewed by HCE establishing that Flock itself has caused Americans to stop going to church, attending protests or participating in political meetings.
That would go beyond the evidence.
The concern is narrower and more fundamental.
A surveillance network does not have to prohibit lawful activity to change the atmosphere in which people exercise it.
The First Amendment protects the right to speak.
Civic life also depends on the willingness to show up.
THREE CAMERAS IN HOOD COUNTY
Hood County did not invent any of these questions.
Its three cameras will not settle them.
But the county offers a remarkably compact version of the argument.
Commissioners approved an expansion of the county’s Flock use in March 2025 with state grant funding. Within months, the court was wrestling with payment, installation, oversight and privacy questions; county officials reported that roughly $132,000 had been paid in connection with a planned 24-camera expansion while only one of those cameras had been installed. (TxDMV)
The dispute continued.
By April 2026, commissioners had decided to terminate the arrangement and seek the county’s money back.
By August, three physical units remained.
Open-records material obtained by HCE identifies all three locations. The August 11th Commissioners Court action likewise referred to three contracted camera units.
Commissioners voted unanimously among the four members present to authorize Shirley’s office to disconnect and place temporary non-damaging coverings over them until their physical removal.
His deputies did.
There was no smashed equipment and no unilateral seizure.
The hardware is physically owned by Flock. Shirley said he did not believe termination of the county’s agreement gave him authority to dismantle, seize or dispose of someone else’s property.
So his office stopped where its authority stopped.
Disconnect.
Cover.
Preserve.
Wait for removal.
The cameras are now inert, Shirley said.
He has not contacted Flock personally, and Flock had not contacted him when HCE interviewed him.
Flock acknowledged receiving detailed questions from HCE on Monday afternoon concerning Hood County’s terminated arrangement, ownership and removal of the equipment, the county’s requested reimbursement, the number of camera units and the company’s broader safeguards. HCE requested a substantive response by 3 p.m. CDT.
Whatever ultimately happens to three pieces of roadside hardware, the larger network continues.
Flock says its systems help find missing people, recover stolen vehicles and identify suspects in serious crimes. It argues that customers control their own information and that stronger audit, retention and access controls can preserve the investigative value of ALPR technology while protecting privacy. Its newly announced reforms substantially strengthen several of those controls. (AP News)
Those are serious arguments.
So are the documented abuses.
So are the erroneous stops.
So are the cybersecurity findings.
So is the possibility that a system purchased for one purpose can later be queried for another.
And so is the problem created when government’s practical ability to reconstruct where people have traveled advances faster than the institutions charged with deciding when government should be allowed to know it.
Shirley believes the answer is not to pretend technology can be stopped.
It is to establish the rules before dependence on the technology makes meaningful limits politically harder to impose.
Warrants where appropriate.
Short retention periods.
Search restrictions.
Meaningful audits.
Protection against personal searches.
Human verification before high-risk enforcement action.
Special protection for constitutionally sensitive activity.
Legislatures willing to draw lines beyond whatever minimum courts eventually require.
If courts ultimately decide that broad ALPR surveillance is constitutional, Shirley said, he will accept the ruling as controlling law.
But that would not end the debate for him.
“Constitutional does not necessarily mean optimal public policy.”
States may protect privacy beyond the federal constitutional floor. Counties and cities may adopt stricter policies. Legislatures may require warrants or limits that courts themselves do not require.
And voters may decide they do not want a particular surveillance system at all.
Shirley sees a distinction between the authority he exercises as a peace officer and the argument he is entitled to make as an elected citizen.
When the law gives him a duty, he says he performs it.
When he believes the law or policy is inadequate, his remedy is political.
That distinction matters because surveillance is ultimately an argument about government power — including the power wielded by the people who oppose surveillance.
A constable should not get to destroy government equipment merely because he dislikes it.
A police chief should not get unlimited surveillance power merely because the technology catches criminals.
Both propositions can be true.
Shirley sometimes describes the role of an elected official in moments like this as standing “in the gap” while institutions catch up to circumstances they were not built to anticipate.
There is danger in that idea, too.
An official convinced that he alone stands between citizens and government overreach can become another form of government overreach.
Shirley’s answer is that the guardrail on him is the same one he wants around the technology:
The law.
That is why the August 11th vote matters.
Commissioners decided.
He acted.
Had they decided differently, Shirley says he would have acted differently.
The cameras would still be watching traffic in Hood County.
Instead, three roadside devices are standing beneath the Texas sky unable to see anything at all.
The country around them is still deciding how much its machines should be allowed to remember.
And how much liberty Americans are willing to surrender for an answer delivered in seconds.
Asked what he hopes people remember when his time as constable is eventually over, Shirley did not mention Flock.
He did not mention cameras.
“That I made Hood County a better place than I found it when I was given this responsibility,” he said.
Whether shutting down three cameras helped accomplish that is a judgment that belongs to the people who gave him the responsibility in the first place.
Disclosure: Shirley is HCE’s Editor-in-Chief. His quotations and firsthand account were verified against the reporting record; the legal, technical and policy reporting was sourced separately. Flock was asked to comment before publication.
