The Unblinking Network
What the Flock is Going On
There is a camera on a pole near your house. It is probably solar-powered, mounted about 10 feet up, angled at the road. It photographs the vehicles that pass through its field of view. It records the plate, the make, the model, the color, the roof rack, the dent in the rear quarter panel, the bumper sticker about your kid’s honor roll status. It logs the time and the location. It uploads all of it to a searchable database. It does this to you, to your neighbors, to everyone, all day, every day, whether or not anyone anywhere suspects anyone of anything.
The company that built it is called Flock Safety. It was founded in 2017, is headquartered in Atlanta, and by mid-2025 carried a valuation of roughly 7.5 billion dollars. Its cameras operate in more than 6,000 communities. Its network includes more than 80,000 automated license plate readers scanning an estimated 20 billion vehicles per month. Roughly 5,000 law enforcement agencies can query its data. So can about 1,000 corporate customers, including FedEx and Lowe’s, plus homeowners associations and private neighborhoods that bought their own cameras and plugged them into the same system.
Flock’s pitch is simple: crime is bad, evidence solves crime, cameras produce evidence; therefore, cameras. The pitch works. It has worked in thousands of city council meetings, usually delivered in 15 minutes with a slide about a recovered stolen car and a slide about an Amber Alert. What does not fit in 15 minutes is the record this company and its customers have actually compiled: the constitutional litigation, the federal agencies slipping through side doors, the officers stalking their ex-wives, the abortion manhunt, the innocent people held at gunpoint over misread plates, stale hotlists, and botched data entry, the exposed camera feeds, the internal debates about buying hacked data, and the aborted plan to wire the whole thing into 27 percent of American doorbells.
This piece assembles that record. It is long because the record is long.
What the Cameras Get Right
Honesty requires starting with what works, because some of it does.
License plate readers recover stolen vehicles. This is not disputed by anyone serious. A camera that alerts on a plate reported stolen 20 minutes ago gives a patrol officer something no amount of shoe leather provides. Flock says it helped locate more than 10,000 missing persons in a single year and supported more than 1 million criminal investigations. Those are company numbers, and company numbers deserve scrutiny, but the underlying mechanism is real. Federal court files reviewed by Forbes show Flock cameras helping catch an alleged Oklahoma drug dealer and an alleged Michigan bank robber. Local news archives hold hundreds of cases where a plate hit closed an investigation that would otherwise have gone cold: hit-and-runs, kidnappings, armed robberies. In Illinois, a Flock hit reportedly helped police apprehend a man who assaulted a 9-year-old girl.
There is also at least one study, commissioned by Flock and conducted with outside academics, finding that adding 1 Flock camera per sworn officer was associated with a 9.1 percent increase in crime clearance rates across 123 agencies. Take that with the salt any vendor-funded study deserves, and more salt arrives below, but clearance is the plausible benefit. Detectives with more evidence close more cases. Nobody should pretend otherwise.
The technology is also cheap by policing standards. A Flock camera runs about 3,000 dollars per year. A detective costs 30 times that. For a small department with 12 officers and a growing caseload of retail theft and stolen Kias, the math on the sales sheet is attractive. That is why the cameras spread so fast: not because chiefs are villains, but because the product is priced and packaged for a 5-minute procurement decision.
All of that is the strongest version of the case for Flock, and it should be stated plainly because the rest of this piece is about what that case leaves out. The question was never whether a national vehicle-tracking database would occasionally catch criminals. Of course it does. A warrantless search of every home in America would also catch criminals. The question is what a free society pays for the catch, who controls the machine, and what the machine’s operators do when nobody is watching them. On that question, the record is now extensive, and it is bad.
The Scale of the Thing
Start with what the system actually is, because Flock’s marketing works hard to make it sound smaller than it is.
A single Flock camera is a still camera on a pole. Flock’s lawyers lean on this. The cameras take “discrete, point-in-time observations,” the company’s Fourth Amendment white paper says, “rather than an always-on dossier of anyone’s movements.” That description is accurate for 1 camera and misleading for 80,000. The product Flock sells is not a camera. It is the network. Each customer’s cameras feed a database, and Flock built sharing features that stitch those databases together: 1-to-1 sharing agreements between agencies, statewide lookup pools, and a “National Lookup” feature that lets a detective in one town query cameras across the country. The company itself has boasted about the reach. A search run by a single Texas deputy in May 2025 touched 6,809 separate networks containing 83,345 cameras. One officer, one search box, most of a continent.
The database is not just plates. Flock’s software fingerprints vehicles: make, model, color, body type, aftermarket features, decals, bumper stickers. An officer who saw “a gray Honda with a Baby on Board sticker” can search for exactly that, no plate required. Flock calls this “Vehicle Fingerprint.” The Cato Institute, in litigation support, called it what it is: a system that lets officers chart a person’s movements across a city with almost no restrictions or oversight. Data is typically retained for 30 days, 21 in states like Virginia that legislated a limit, and the window rolls forward forever. Any 30-day slice of your driving life is available to any of roughly 5,000 agencies with the stated “reason” field filled in. The reason field is a text box. Officers type a few words. Nobody approves the search before it runs.
Flock’s answer to every governance question is that customers control their own data. “Each Flock customer has sole authority over if, when, and with whom information is shared,” the company told NPR. The record shows what that authority is worth. City after city discovered, after the fact, that their data was flowing to agencies they never approved: Border Patrol accounts querying local cameras, out-of-state departments running millions of searches, federal task force agents borrowing local logins. Auburn, Washington, announced that federal access to its system “occurred unknowingly to us.” Lakewood’s police chief said Border Patrol “somehow got temporary access to a Flock account” without the department’s knowledge. When the customer with “sole authority” cannot tell you who searched its cameras last month, the authority is decorative.
And the network keeps growing appetites. Flock acquired a drone company. It launched Raven, an audio detection product marketed for gunshots that the company acknowledged can detect human voices, prompting the Electronic Frontier Foundation to warn about always-listening infrastructure. It built Nova, a people-search platform discussed below, designed to let an officer “jump from LPR to person” and map that person’s associates. The camera on the pole was never the product. It was the on-ramp.
The Fourth Amendment Problem
The Fourth Amendment protects the right of the people to be secure against unreasonable searches. For most of American history, practical limits did the heavy lifting. Police could always watch a car on a public road, and courts said so: United States v. Knotts, in 1983, held that a driver on public thoroughfares has no reasonable expectation of privacy in his movements, because anyone could watch him. That logic assumed watching was expensive. An officer can follow one car. He cannot follow every car forever, and remember all of it.
Flock removed the practical limits, and the courts are now fighting over whether the legal ones survive. The key precedent is Carpenter v. United States, the 2018 Supreme Court decision holding that police need a warrant to obtain historical cell-site location records, because pervasive, retrospective location tracking invades a reasonable expectation of privacy even in public movements. The Fourth Circuit applied similar logic in Leaders of a Beautiful Struggle v. Baltimore, striking down a program that photographed the whole city from the air. The question courts keep facing: is a dense network of AI plate readers more like one cop watching one road, or more like Carpenter’s dragnet?
The test case is Norfolk, Virginia, a city of about 230,000 that installed 172 Flock cameras, roughly 1 for every 1,300 residents. Two residents, Lee Schmidt and Crystal Arrington, sued in October 2024 with the Institute for Justice after discovering the cameras had photographed their vehicles 475 and 325 times, respectively, over 4 months. Their complaint put it plainly: the city lives under a flock of unblinking eyes that never sleep and remember everything. In February 2025, a federal judge refused to dismiss the case, finding the allegations “notably similar” to Carpenter and that warrantless queries of the database plausibly violated the Fourth Amendment. Around the same time, a Norfolk circuit judge in Commonwealth v. Bell suppressed evidence from a warrantless Flock search in a criminal case.
Then the tide ran the other way. In October 2025, the Virginia Court of Appeals reversed the same judge in Commonwealth v. Church, holding that no warrant was required because the system “simply took pictures” of a car on public roads. In January 2026, the federal court granted Norfolk summary judgment in Schmidt, reasoning that 176 cameras on a 21-day rolling window “does not track the whole of a person’s movements” nor provide an “intimate window” into where citizens “drive, park, visit, linger, sleep, or patronize.” Flock published a victory blog the same day and reminded customers that more than 30 state and federal courts had reached similar conclusions, including in United States v. Martin, where a Virginia federal court found no privacy interest because 188 cameras captured only 3 photos of the defendant’s car in 30 days.
Read those holdings carefully and notice what they actually say. The courts did not hold that mass vehicle tracking is fine. They held that this particular deployment had not yet crossed the Carpenter line: not enough cameras, not enough retention, not enough captures of this particular plaintiff. The Martin court said it directly, acknowledging the defendant’s warning that camera proliferation and database sharing would eventually create the forbidden dragnet, and answering only that “the future is uncertain” and the facts were not there yet. That is not a constitutional endorsement. That is a court measuring how much of the noose has been tied.
Flock’s own defense concedes the frame. Its white paper argues its deployments are constitutional because of the 30-day retention limit, the fixed camera locations, and the limited capture density. Every one of those safeguards is a product setting. Flock controls them, and Flock’s business model is to increase camera density in every market it serves, to encourage nationwide sharing, and to sell add-ons that link plates to people. The company is litigating on the theory that its network is too sparse to be unconstitutional, while selling on the theory that its network sees everything. Both cannot stay true.
And in June 2026, the ground shifted. The Supreme Court decided Chatrie v. United States, holding 6 to 3, in an opinion by Justice Kagan, that police conduct a Fourth Amendment search when they obtain a person’s cell-phone location data, and rejecting the government’s argument that pulling only a narrow, time-limited slice of a giant database avoids the problem. Once the Fourth Amendment applies, the size of the bite does not matter. Chatrie involved cell-phone data, not plate readers, and does not decide the Flock question by itself. But the plaintiffs and their amici argue, with some force, that its reasoning cuts directly against the argument Norfolk won on: that 21 days and 176 cameras is a small enough slice. Schmidt is now on appeal at the Fourth Circuit as No. 26-1227, with the ACLU, ACLU of Virginia, and EFF filing a joint amicus brief in April 2026 arguing that Flock’s networked databases let agencies run retrospective surveillance on people suspected of nothing. The states defending Norfolk had leaned on the vacated lower-court version of Chatrie. Whatever the Fourth Circuit decides, that crutch is gone.
Meanwhile, the state constitutional front opened. In November 2025, EFF and the ACLU of Northern California sued San Jose, alleging its Flock program violates the California Constitution and state law. The numbers in that case deserve their own sentence: between June 5, 2024 and June 17, 2025, San Jose’s database was searched 3,965,519 times by SJPD and other agencies, warrantless, every one. Separate reporting found SFPD allowed out-of-state agencies to run more than 1.6 million searches of its Flock data that reporters and civil liberties groups said were barred by California law, including at least 19 flagged as ICE-related. A class action filed in 2026 alleges a New York county scanned residents’ plates without authorization. The Institute for Justice’s Plate Privacy Project is running parallel litigation and legislative campaigns nationwide.
The counterargument, made by Flock’s chief legal officer Dan Haley and by dozens of courts, is that a license plate is public by design, a government-issued identifier displayed on a public road, and photographing it is no different from an officer jotting it down. True, as far as it goes, and it goes about 1 camera’s worth of distance. An officer jotting down a plate cannot reconstruct where a woman drove for the last 3 weeks, cross-reference it against 82 other cities, and do it 4 million times a year without anyone signing a warrant. The Fourth Amendment was written by people who had experienced general warrants: blanket authority to search everyone in hopes of finding someone. A database of everyone’s movements, queryable on an officer’s say-so, is a general warrant with better resolution. The founders did not fail to ban it because they approved of it. They failed to ban it by name because they could not imagine anyone building it.
The People Holding the Keys
Every surveillance system is an argument about human nature. Flock’s argument is that roughly 140,000 monthly law enforcement users, given warrantless access to the movements of nearly every driver in America, will restrain themselves because a text box asks them why they are searching. The evidence is in, and the argument lost.
Start with the stalking. The Institute for Justice, reviewing media reports and public records, has documented at least 21 cases of police officers using license plate reader networks to track people they were romantically interested in: current partners, exes, rivals, and in at least one case a stranger an officer spotted in public and decided to find. Most of the cases have accumulated since 2024, which is to say since the network got dense enough to be useful for the purpose. The roster reads like a personnel file from a department nobody should fund. In Kechi, Kansas, Lieutenant Victor Heiar pleaded guilty to computer crime and stalking after using Flock cameras to track his estranged wife. In Sedgwick, Kansas, Police Chief Lee Nygaard ran his ex-girlfriend’s plate 164 times and her new boyfriend’s 64 times over 4 months before resigning. A Jerome County, Idaho sheriff ran his wife’s plate more than 700 times in 3 months, labeling each search “test,” then retired. In Orange City, Florida, Officer Jarmarus Brown ran his girlfriend’s plates 69 times, her mother’s 24 times, and her brother’s 15 times while also slipping an AirTag into her wallet; he got probation. Costa Mesa, California: Officer Robert Josett pleaded guilty after tracking his mistress and her other romantic interests. Riverside County: a deputy arrested for kidnapping his ex-fiancée had used the system to track her friend. Louisville: felony charges. Kenosha County: a deputy resigned with severance after tracking a coworker. Braselton, Georgia: the police chief himself was arrested in November 2025 for allegedly using the readers to stalk and harass multiple people, including a former partner. Holiday Hills, Illinois: another chief, charged in June 2026 with felony official misconduct for tracking 6 people he knew personally, 3 of them women he had dated. Monroe County, Florida, 2026: a deputy allegedly used the system to track and then pull over a woman he met while working security on a TV set.
Flock’s response to this pattern is that abuse is “rare” among its users and that comprehensive audit trails provide accountability. Examine how these cases actually surfaced and the audit trail defense collapses. Only a handful of the 21 were caught by internal review. Most came to light because the victim felt watched, filed a stalking complaint, and forced an investigation that then found the searches. The audit log did not protect anyone. It documented the harm after a frightened woman did the detective work herself. A log that fires no alarm is not a safeguard. It is a diary. Some victims discovered the surveillance through HaveIBeenFlocked.com, a volunteer-run site that compiles public search records, which means a hobbyist website currently provides more proactive oversight of America’s largest plate-tracking network than the network’s operators do. The fix here is not mysterious: mandatory automated flagging of repeat searches on a single plate, independent audit with penalties, and a warrant requirement that puts a judge between an officer’s curiosity and a citizen’s location history. Departments that will not implement those controls have told you what the tool is for.
The abuse is not limited to romance. In Lenexa, Kansas, a resident named Canyen Ashworth wrote a column critical of the city’s police. A KCUR investigation found the department subsequently tracked his car through its plate readers and opened a criminal investigation suggesting he was the person illegally posting memes near city hall, complete with an internal be-on-the-lookout notice. He was not the man in the video. Nobody was charged. The department says the investigation is inactive but could be reopened. A man criticized his police department in print, and the department’s response was to run his movements through a surveillance network. That is not a hypothetical chilling effect from a law review article. That is a Kansas suburb, on the record.
Then there is the profiling. EFF’s analysis of more than 12 million logged searches across 3,900 agencies found more than 80 law enforcement agencies using slurs and stereotypes against Romani people in their search reasons, including “possible gpsy” and “gpsy ruse.” One Texas department ran the slur 6 times through Flock’s “Convoy” feature, which identifies vehicles traveling together, targeting an entire traveling community with no crime specified. The reason field, remember, is the entire pre-search oversight mechanism. This is what officers type into it when they believe nobody is reading.
And the protests. Through 10 months of nationwide audit logs, EFF documented more than 50 federal, state, and local agencies running hundreds of Flock searches connected to protest activity: the 50501 protests in February 2025, the Hands Off protests in April, the No Kings protests in June and October. Nineteen agencies ran dozens of searches tied to No Kings demonstrations alone. Tulsa police logged at least 38 protest-related searches, including one corresponding to a demonstration supporting a detained pro-Palestinian activist. Delaware State Police queried the network 9 times over actions by an animal-rights group. Sometimes the logs said “protest” outright. Sometimes agencies used vaguer language, which matters because Flock has advised agencies to be “as vague as permissible” in their audit log entries. The company that markets audit logs as its accountability backbone coaches its customers on how to make the logs useless. Sit with that for a moment before reading the next section, because the next section is about what happens when the customers include the federal government.
The Federal Side Doors
Flock spent years telling city councils a simple story: your cameras, your data, your rules. No federal contracts. Local control. In 2025, that story came apart in public, piece by piece, and the way it came apart matters as much as the facts themselves.
In May 2025, 404 Media obtained search logs showing that local and state police around the country had run more than 4,000 nationwide or statewide Flock lookups for immigration purposes, either at the request of federal agencies or as informal favors. ICE had no Flock contract. It did not need one. It asked local cops, and local cops typed “immigration” or “ICE” into the reason field and searched cameras across the country. The University of Washington’s Center for Human Rights later mapped the anatomy in Washington state: “front door” access, where at least 8 local agencies enabled direct 1-to-1 sharing with Border Patrol; “back door” access, where Border Patrol reached data from at least 10 agencies that never authorized it; and “side door” access, where officers ran searches on federal agents’ behalf. A Renton audit showed dozens of searches of the local network by agencies around the country citing “ICE” and “immigration” as the reason. Only 3 Washington jurisdictions showed no exposure at all, and they shared a trait: they had restricted access to in-state agencies only. Governance worked exactly where it existed and nowhere else.
Then Colorado journalists found the contract-shaped hole. In August 2025, 9News revealed that Customs and Border Protection had direct access to Flock’s backend through a “pilot program” the public had never heard of. Days later, data released by 3 police departments showed CBP had been regularly searching more than 80,000 Flock cameras, functionally the entire national network. One of the departments told 404 Media it did not know or understand that it was sharing data with CBP. Flock CEO Garrett Langley had previously denied that federal contracts existed. He then issued a statement acknowledging that “some of our public statements inadvertently provided inaccurate information,” that Flock had run limited pilots with CBP and Homeland Security Investigations, and that the company “didn’t create distinct permissions and protocols in the Flock system to ensure local compliance for federal agency users.” Strip the corporate passive voice and what remains is Flock’s own admission: the country’s largest surveillance vendor acknowledged that its public statements about federal access to its network had been inaccurate, and the accurate version surfaced only because local reporters and anonymous researchers pried loose the logs. Flock says the pilots have been paused. A pause is a business decision. It reverses when attention moves.
The state-law problems were not hypothetical. Illinois passed a law in 2023 prohibiting license plate reader data from being shared for immigration enforcement or abortion investigations. Secretary of State Alexi Giannoulias audited Flock after the abortion search described below, sampling just 12 local agencies, and found CBP accessing Illinois camera data anyway. He publicly stated that Flock had no proper safeguards for data sharing and was running the CBP pilot in violation of state law, and ordered access cut. His office noted a detail that should end careers: Flock leadership was reportedly unaware of the CBP pilot running inside its own product. In Oak Park, early audit work found a Flock account tied to Palos Heights police searching Oak Park data 28 times for “immigration violation.” Investigative outlet Unraveled then reported the mechanism: a Palos Heights detective had shared his Flock login with a DEA task force agent, who used it to run unauthorized immigration searches of Illinois’s database. When the detective mentioned in a group chat that his login had been changed because of those searches, the agent responded with a sitcom GIF. That is the operational security culture guarding the movements of 12 million Illinois drivers. Notably, Illinois’s law contains no specific penalties for violating it, which is why the violation cost Flock nothing but a press cycle. A law without a penalty is a suggestion, and legislators who write suggestions should stop taking credit for writing laws.
The pattern repeated wherever anyone looked. Dayton, Ohio, found more than 7,000 immigration-related searches of its data by outside entities, called them egregious policy violations, and stopped using Flock. San Francisco’s data absorbed 1.6 million out-of-state searches barred by California law. Virginia agencies ran nearly 3,000 immigration-related searches on the state’s network in a single year, per the Virginia Center for Investigative Journalism, and a Richmond ATF analyst improperly used the city’s Flock system for an immigration case, after which Richmond cut federal access. Boulder discovered its cameras had been included in more than 100 Border Patrol searches, and over 4,000 immigration-cited queries flowed through the national network in under a year. A congressional letter from Representatives Garcia and Krishnamoorthi in August 2025 demanded answers about all of it. Flock’s structural response was to remove federal users from statewide and national lookup and to add search filters and an offense-type dropdown menu. A Santa Cruz councilmember identified the flaw in the dropdown fix immediately: an officer running an immigration search can simply select a more palatable menu option. Her city canceled its contract in January 2026. The dropdown remains.
None of this required hacking, warrants, or secret orders. That is the finding that matters. The nation’s immigration enforcement apparatus obtained a de facto national vehicle-tracking capability through checkbox settings, borrowed logins, informal favors, and a pilot program the vendor’s own leadership claims not to have known about. Every safeguard was a policy, every policy was voluntary, and every voluntary policy failed on contact with an agency that wanted the data. Defenders respond that data sharing between law enforcement agencies is normal and often good, and within a framework of warrants and jurisdiction, it is. This was the opposite of a framework. It was a market: local agencies held the inventory, federal agencies held the demand, and Flock’s system supplied the exchange even as the company’s public messaging emphasized customer-controlled sharing.
The Abortion Search
On May 9, 2025, an officer from the Johnson County Sheriff’s Office in Texas ran a search across Flock’s national network. The logs show it reached 6,809 networks and 83,345 cameras, including cameras in Washington and Illinois, states where abortion is legal and protected. The reason field read: “had an abortion, search for female.”
When 404 Media broke the story, Sheriff Adam King said the search had nothing to do with criminal enforcement. The woman had self-administered an abortion, he said, her family feared she was bleeding to death, and deputies were trying to get her to a hospital. Flock backed him in writing: “She was never under criminal investigation by Johnson County. She was being searched for as a missing person, not as a suspect of a crime.” The company accused journalists and EFF of spreading clickbait.
Then EFF obtained the court records. The responding deputies had opened a “death investigation” into a “non-viable fetus.” They collected and logged evidence of the woman’s self-managed abortion: photographs, the FedEx envelope the medication arrived in, the administration instructions, all of it assembled by her partner, who reported her to police and who would later be charged with domestic violence against her. Investigators consulted the district attorney’s office about charging her and were told they could not. The Flock searches, 2 of them, ran while that death investigation was open. When the woman later walked into the sheriff’s office, deputies interviewed her about the fetus, inspected the abortion-related texts on her phone, and had her write a timeline of events. Only afterward did they learn why she had actually come: to report that less than an hour after her abortion, her partner choked her, put a gun to her head, and made her beg for her life. He was charged. The sheriff, separately and later, was indicted on felony counts including aggravated perjury in an unrelated whistleblower retaliation case. That is the man whose word Flock staked its public credibility on.
Consider what the episode establishes, point by point. First, capability: a single deputy in a state where abortion is banned searched cameras in states where it is a protected right, in seconds, with no warrant, no court, and no notice to the jurisdictions searched. Officials in Spokane County, Washington, learned their cameras were included only from records requests months later; the county’s initial public statements did not even acknowledge it. Second, intent: the search reason was written by the officer himself, and it was not a “welfare check.” Third, the vendor’s public response: Flock’s initial statements repeated the sheriff’s welfare-check account and criticized the reporting as clickbait, and the sworn affidavit of the responding investigator, obtained later by EFF, contradicted that account. If/When/How’s research found that about 26 percent of adult criminal cases over self-managed abortion begin when a friend, parent, or intimate partner reports the person to police. Flock built the machine that turns that phone call into a nationwide manhunt. The company’s defenders note that courts have repeatedly protected the right to interstate travel, which is true and beside the point. Rights are not protected by their existence on paper. They are protected by the absence of infrastructure for violating them at scale. The infrastructure now exists, it is subscription-priced, and it has been used.
Wrong Car, Guns Out
The civil liberties argument against Flock is sometimes framed as abstract, a matter of principle for people who have nothing to hide. The people below had nothing to hide. Several of them had guns pointed at them anyway.
The Institute for Justice reviewed media reports and court records and found at least 24 cases since 2018 of innocent motorists pulled over, detained at gunpoint, or jailed because of stationary license plate reader errors, most since 2023. The list excludes mobile readers and toll cameras, and IJ calls it an undercount, likely a significant one, because almost nobody tracks these failures. In February 2026, a Flock camera in Sherwood, Arkansas, misread a plate digit, and officers detained an innocent couple at gunpoint while their 6-week-old baby sat alone in a car seat. The officer’s apology as he uncuffed them deserves preservation: “I’m not gonna say they’re completely perfect, because, you know, that’s modern technology.” Two sisters, gunpoint, misread digit. A teenager, gunpoint, bad data entry. A woman was detained at gunpoint while her autistic child was placed in the back of a police car and her vehicle was impounded for weeks. A driver was jailed 13 days after officers mistook his car for one at a deadly accident. Another was jailed for nearly a month over an attempted carjacking he had nothing to do with. Multiple Colorado drivers were pulled over again and again because their plates were mistakenly left on hotlists nobody bothered to clean.
The failure modes compound because the network is national. In June 2026, automotive journalist Joel Feder was test-driving a Range Rover press car in suburban Minnesota when 4 squad cars boxed him in outside a Kohl’s, officers shouting with hands on their guns. Plymouth police had been tracking him for days through Flock cameras, waiting for the right moment. The trail began 2,000 miles away: a Los Angeles dealership reported plates missing, the plate was entered into the system without its small-format middle digits, and Flock’s AI began flagging every Jaguar Land Rover press vehicle with the same plate structure, nationwide, as stolen. The plates were not even stolen; they had been misplaced during a photo shoot. Officers warned Feder to drive straight home and avoid neighboring towns, whose cameras would flag him all over again. “You’re lucky we’re in Plymouth,” one said. “If you were in Minneapolis, they definitely would’ve come at you with guns drawn.” One clerical error, propagated instantly to every jurisdiction in America, self-refreshing, correctable nowhere. That is not a safety architecture. That is a rumor mill with police powers.
The errors do not need to be dramatic to be corrosive. In Columbine Valley, Colorado, police issued Chrisanna Elser a criminal summons for stealing a package based solely on Flock images of her truck passing through town. She spent weeks assembling her own defense: timestamped video from her truck, a neighbor’s doorbell footage, proof she was never at the scene. Police voided the summons after she did their investigation for them. No policy required the officer to verify the automated hit before charging her. The burden of proof migrated, quietly, from the state to the citizen, and it migrated because the machine’s output looks like evidence and paperwork is easier than police work.
How often does this happen? Almost no jurisdiction measures it, which is itself the scandal. IJ found exactly 1 that publishes regular error reports: Oak Park, Illinois. Those reports show that in a typical period, one-third or more of traffic stops prompted by a Flock alert ended with the driver released because of data problems. One-third. That figure measures the stops, not the cameras; the failures in those reports trace variously to misreads, stale hotlists, and entry errors upstream of the camera. But the driver on the curb does not experience the distinction, and Flock’s marketing calls the system an evidence machine. In the one town that counts, at least a third of the stops the machine prompted ended with police letting an innocent driver go. The remedy is standard practice from any field that takes error seriously: mandatory human verification of the physical plate before a stop, mandatory error logging, public error reporting, and liability when an unverified machine hit puts a gun in a family’s face. Departments adopted none of it by default, because the vendor sells confidence and nobody was assigned to sell doubt.
The Security of the Watchers
A company that photographs everyone’s movements 20 billion times a month carries an obligation that precedes every other promise: keep the archive locked. Flock’s record on its own security is a case study in the gap between marketing and engineering.
In late 2025 and January 2026, researchers and reporters in Florida found that Flock’s Condor PTZ cameras, the pan-tilt-zoom video units police use to track people and objects with AI, had feeds exposed on the open internet. Anyone who found the links could watch live video, review roughly a month of archived footage, and even delete video from the system. More than 60 exposed Condor units were identified in one round of research, some overlooking places like a children’s playground. The musician and researcher Benn Jordan, who documented the exposure, described what the archives revealed: patterns of daily life, like the fact that a particular woman walked alone on a particular forest trail at 7 p.m. He called it Netflix for stalkers. Flock confirmed the incident to the local press and blamed a testing configuration on a small number of units. The same company’s public FAQ, published weeks later, is titled “Has Flock Been Hacked?” and answers no, no breach, no compromise, ever. Both statements are technically compatible, and only one of them is informative. Unauthorized strangers watched and could delete police surveillance footage. Whether the dictionary calls that a breach is a question for the marketing department, not for the woman on the trail.
The pattern extends backward. In May 2025, Flock publicly responded to disclosed vulnerabilities in its license plate readers and gunshot detection devices. In November 2025, an independent researcher published a white paper compiling security findings across Flock’s hardware and software; Flock registered CVEs and issued customer advisories while stressing the flaws were theoretical and required physical access. Earlier research described a misconfigured Flock demo environment that exposed the scope of its 83,000-camera network. In December 2025, Flock appointed its first executive carrying the specific title of chief information security officer. The company had security leadership before that, and the appointment is a fine step. The date is still worth noticing: the firm was 8 years old, valued at 7.5 billion dollars, and operating the largest vehicle surveillance network in American history before the industry-standard title for guarding such a network appeared on its org chart.
Then there is Nova, the product that shows where the road goes. Nova is Flock’s people-search platform, built to let investigators, in a Flock employee’s own recorded words, “jump from LPR to person and understand what that context is, link to other people that are related to that person, marriage or through gang affiliation, et cetera. There’s very powerful linking.” In May 2025, 404 Media reported internal discussions about powering Nova partly with breached data, including records from the 2021 ParkMobile hack. Flock’s own employees objected internally; one wrote he was “pretty horrified to hear we use stolen data in our system,” noting it could create perverse incentives for more data to be stolen. After the story broke, Flock announced at an all-hands and in a blog post that Nova would not supply dark web data, framing the reporting as premature and based on dubious sources. In December 2025, an outside security researcher analyzing Nova’s front-end code reported a data source labeled “Dark Data,” with search selectors for Social Security numbers, credit card numbers, crypto wallets, and messaging handles, and table columns for leak name, leak host, and download location, the vocabulary of breach archives. Code is not proof of deployment. Internal discussions are not proof that breached data ever reached a customer, and Flock’s stated policy, published in May 2025, is that Nova does not and will not supply data from breaches. The operational reality remains disputed, and the researcher’s specific findings have not been reconciled point by point in public. What can be said without dispute is narrower and still damning: the company explored sourcing breach data, by its own account, and stopped only after its employees objected and a reporter called. The burden of clarity now belongs to a firm whose statements about federal access had to be corrected, whose account of the abortion search was contradicted by court records, and which described its exposed camera feeds as not-a-breach. Trust is a budget. Flock has spent its.
The Doorbell Play
In October 2025, Flock announced a partnership with Ring, Amazon’s doorbell camera division and the most popular residential camera brand in the country. Roughly 27 percent of American households now have a doorbell camera. The integration would have connected law enforcement agencies using Flock’s Nova platform and FlockOS to Ring’s Community Requests feature, letting police solicit doorbell footage from residents through Flock’s own software. Requests would include a location, timeframe, and investigation code; sharing would be voluntary and anonymous, both companies stressed.
Understand what the combined system would have been. Flock’s cameras track the vehicle to the block. Ring’s cameras capture the face at the door. Ring has deployed facial recognition features, including “Familiar Faces,” which scans and matches the faces of people within camera view. A plate scan seeding a request for face-level footage from private porches, brokered through the same police platform that already ingests plate data, jail records, computer-aided dispatch, and open-source intelligence, is not 2 products. It is 1 surveillance stack, reaching from the public road to the front step, and the privacy experts who said so at the time were not speculating. Stephen Perez of Restore the Fourth pointed to the operative clause: Flock’s default agreement lets departments share data with federal and local agencies for investigative purposes even when a local department restricts its own data, making it a “very real possibility” that footage volunteered for one local case would flow to ICE or HSI for another. Jon Gaines, the researcher who had documented dozens of Flock vulnerabilities, made the security point: bolting the country’s biggest residential camera fleet onto a company with Flock’s history of misconfigurations and access-control failures multiplies the attack surface for everyone.
Ring’s own history sharpened the concern. The company had previously handed footage to police without warrants or owner consent at least 11 times, then in 2024 announced it would require warrants, a retreat from years of police-friendly design. The Flock deal was a return trip. Then, in February 2026, Ring aired a Super Bowl ad for a feature called Search Party, showing neighborhood cameras pooling AI to track a lost dog through the streets. The public watched a network of private cameras collaboratively tracking a moving target and drew the obvious conclusion about what else the target could be. The backlash was immediate; Senator Ed Markey called the technology creepy in an open letter; EFF noted the facial recognition already built into the product line. Within the week, Ring and Flock announced a “joint decision” to cancel the integration, which had never gone live. Ring’s stated reason was that the project “would require significantly more time and resources than anticipated,” a sentence that has never once in corporate history meant what it says. No Ring footage ever reached Flock. That is the good news, and its shape is instructive: the pipeline from America’s doorbells into a warrantless national tracking platform was stopped not by a court, not by a legislature, not by either company’s ethics review, but by 30 seconds of bad publicity during a football game. Amazon’s Community Requests program continues through its partnership with Axon. The idea is not dead. It is waiting for a quieter news cycle.
Does Any of This Even Reduce Crime?
Set aside every abuse above and grade Flock on its own terms: does blanketing a town in plate readers make the town safer? The honest answer is that nobody has proven it, including Flock, and Flock has tried hard.
The company’s flagship claim, from a white paper written by 2 Flock employees and 2 academics, is that 10 percent of reported crime in the United States is solved using Flock technology. Forbes put that claim to 6 independent criminal justice academics. Their reaction ranged from problematic to, in the words of University of Texas sociologist Michael Sierra-Arevalo, bordering on ludicrous absent a four-alarm research finding. Forbes also checked the ground truth in a small city Flock had showcased: the company installed cameras and claimed crime went down, and crime went up. In Lexington, Kentucky, where crime dipped only slightly after Flock’s 2022 arrival, the police department’s own lieutenant told Forbes he did not believe the technology would have a demonstrable impact on crime rates. The clearance-rate study mentioned at the top of this piece, the one finding a 9.1 percent improvement, was likewise Flock-commissioned, and clearance is not crime reduction; a department can clear more cases in a town where crime is flat or rising. Flock’s 2026 “Impact Census,” its latest evidence offering, is a survey of its own customers, which the company itself concedes is “a starting point, not a finished product.” Vendors do not get to grade their own homework and then cite the grade in procurement meetings, but that is precisely the market’s current standard of proof.
The deterrence theory has a specific problem: plate readers deter people who believe their plate will be read. Car thieves already steal plates and swap them, a fact Flock’s own marketing acknowledges when explaining why it fingerprints vehicle features. So the system’s deterrent pressure lands mostly on people who are not planning crimes, which is another way of saying it lands on everyone as surveillance and on criminals as a paperwork inconvenience. Meanwhile, the opportunity cost is real money. El Cerrito, California, calculated that ending its Flock contract saves 315,000 dollars over 3 years. Weston, Missouri, was asked to spend 45,000 dollars for 2 cameras. Multiply across 6,000 communities, and the nation is spending serious sums on a technology whose independent evidence base, after 8 years and 80,000 cameras, consists mainly of anecdotes the vendor circulates daily and studies the vendor paid for. Cameras do solve individual cases; the recovered-vehicle stories are real. But a tool that solves some cases while prompting stops, of which at least a third ended in release for data problems in the one town that publishes error reports, enabling stalking by at least 21 officers, and exposing whole populations to federal dragnet searches is not a safety product with side effects. It is a surveillance product with a safety brochure.
The Company’s Character
A pattern this long stops being a series of incidents and becomes a description of the firm. Consider the throughline.
When 404 Media revealed the ICE side-door searches, Flock’s CEO denied federal pilot relationships that the company later confirmed, acknowledging its earlier statements had been inaccurate. When the abortion search surfaced, Flock repeated the sheriff’s welfare-check story and attacked the journalists until the affidavit showed a death investigation. When the Nova reporting broke, Flock called the sources dubious, then held an all-hands to announce it would not do the thing the sources described. When its camera feeds were exposed online, it published an FAQ asserting it had never been breached. When audit logs became a liability, it advised customers to be as vague as permissible in filling them out. When Arizona legislators drafted a plate-reader bill in 2026, it included a section banning public records requests for the data, the exact mechanism through which every abuse in this article came to light. And when a volunteer open-source project called DeFlock began mapping camera locations so citizens could see the network built around them, Flock’s CEO publicly likened the project to terrorism, a characterization captured on video and reported by multiple outlets. That is the posture of the company toward the public it photographs: transparency for thee, vagueness for me, and a slur for the people who make maps.
Even the company’s core denial does not survive its own executives. Flock’s communications chief says on the company’s Trust page that “there’s a common misconception that Flock tracks you wherever you go, and that’s just not the case,” because Flock tracks vehicles, not people. Its chief legal officer, on the same page, explains that license plates “are required specifically to correlate to ownership of that vehicle.” IPVM, the security industry’s independent research outlet, put the 2 statements side by side and stated the obvious: a plate read is by design a record tied to a person, and a police chief running a romantic rival’s plate 140 times is tracking a person; the vehicle is the mechanism, the person is the target. The private-sector reach makes the same point from another angle. Flock’s cameras are not only municipal. FedEx, Lowe’s, HOAs, and shopping centers run them and can feed the network. In April 2026, public records reportedly showed Flock employees accessing a private camera network at a Jewish community center in Dunwoody, Georgia, including views of the pool area while children swam. Flock has said the access occurred under an authorized testing and development agreement with the city, and that it is changing its practices around sensitive locations. The report rests largely on a single outlet and the underlying records, and the company’s explanation deserves weight; so does the fact that the company felt compelled to change its practices. A firm with this record asking for the benefit of the doubt is asking for a subsidy.
The Saws Come Out
Predictably, some Americans stopped petitioning and started cutting.
In Suffolk, Virginia, an Air Force engineer named Jeffrey Sovern was arrested in October 2025 and charged with 13 felony counts of destruction of property, plus larceny and burglary-tool counts, over 13 Flock cameras damaged across North Suffolk between April and October. According to the criminal complaint, he told investigators he disassembled the 2-piece poles with vice grips and kept some wiring, batteries, and solar panels; publicly, he has pleaded not guilty and stopped short of admitting the conduct, while stating that plate reader systems are unconstitutional and that Flock is building an unhealthy surveillance state. Privacy advocates nationwide donated more than 15,000 dollars to his defense. The detail that turned a local vandalism case into a national story: the man accused of destroying surveillance cameras was identified by surveillance, a traffic camera catching his truck near a malfunctioning Flock unit, followed by a GPS tracker warrant. In Eugene and Springfield, Oregon, at least 6 cameras were cut from their poles in October 2025, with a note left behind: “Hahaha get wrecked ya surveilling” and a word this publication will let readers reconstruct. In Newport News, a camera pole was destroyed with no suspect found. In Barrow County, Georgia in July 2026, someone took a saw to 3 camera poles and their solar feeds along Double Bridges Road and pushed over a fourth near a high school. Straight Arrow News, analyzing audit logs, found agencies across the country running Flock searches to hunt people who damage Flock cameras, including one Indiana sheriff’s employee who ran 9 searches across roughly 94,000 cameras over a single destroyed unit, and an Ohio sheriff’s office whose logged search reason was “flock camera got shot.”
Be clear about 2 things at once. Cutting down a camera is a crime; it should be, and this publication does not endorse it; a citizen who destroys public property has substituted his own judgment for his neighbors’, which is the same sin the surveillance state commits with better funding. But notice what the state’s response reveals about its priorities. In Georgia, damaging a Flock camera is a felony, announced with a sheriff’s vow of prosecution to the fullest extent of the law. In Illinois, the platform used for searches that state officials said violated the surveillance statute, exposing millions of drivers to federal immigration queries, faced a law with no penalties at all. A man with a saw gets 13 felony counts. The conduct the Secretary of State called a clear violation of state law got a strongly worded audit. Citizens can see the asymmetry, and the asymmetry, not the vandalism, is the real threat to public order. People do not saw down infrastructure they believe they consented to and control. The saws are a symptom. The disease is that in most of these 6,000 communities, nobody asked the residents before the poles went up, and when residents objected afterward, at least one county commissioner told them flatly, on the record, “You will not speak on Flock tonight.”
What Actually Works Against This
The encouraging part of this story is that ordinary communities, acting through ordinary local institutions, have repeatedly beaten a 7.5 billion dollar company, and the playbook is now well documented.
The first tool is the contract itself. Flock’s entire deployment model runs through renewable municipal subscriptions, which means every community holds a kill switch on a schedule. Communities have used it. Austin canceled. Evanston deactivated its cameras and moved to terminate. Oak Park’s board voted 7 to 0 on the territory before the state audit even landed. Dayton quit after finding the 7,000 immigration searches. Santa Cruz voted out in January 2026. Flagstaff’s mayor, who initially wanted guardrails, concluded the community had simply lost trust in Flock and ended the contract. Los Altos ended its contract in January 2026; El Cerrito declined renewal in May and pocketed 315,000 dollars. In Oregon, the Lane County Sheriff suspended, Eugene and Springfield canceled, and Bend declined to renew, all within about 2 months. Columbia Heights, Minnesota, voted to remove every camera. Denver did not even wait for its own scandal; it cut off a neighboring city’s access after that city admitted funneling searches to federal agents. Flock counters that new agencies signing up still outpace cancellations, which is true and beside the point: the cancellations prove the decision is reversible, local, and winnable, which is more than can be said for most surveillance fights. A community that ends a Flock contract has not filed an amicus brief or waited on the Fourth Circuit. It has met, argued, voted, and removed the cameras from its own streets, which is what self-governing communities are for.
The second tool is sunlight, and nearly all of it has come from outside the government. The abuses in this article were surfaced by a short roster: 404 Media’s reporters, EFF’s records requests covering 12 million search logs, the Institute for Justice’s litigation and tallies, university researchers in Washington, local outlets like 9News and Boulder Reporting Lab and RANGE and Unraveled, a Muckrock user filing FOIAs, the volunteers behind DeFlock’s camera map and HaveIBeenFlocked’s search-log archive, and one YouTuber who found police camera feeds sitting on the open internet. The major incidents examined in this article were first brought to light by journalists, researchers, litigants, or outside audits, not by Flock or by the departments’ own internal reviews. That is the civic lesson worth underlining: the oversight layer that actually functions here is voluntary, distributed, and privately organized, and any community that wants to govern this technology should fund and use that layer rather than waiting for the vendor’s transparency portal to confess something. File the records requests. Read the network audits. Check the sharing settings quarterly, because Auburn and Lakewood did not.
DeFlock deserves its own paragraph because it is the clearest working example of that voluntary oversight layer, and because readers can use it today at DeFlock.me. The project was started in October 2024 by Will Freeman, a software engineer who noticed black camera poles with solar panels multiplying along his drive from Washington state to Huntsville, Alabama, photographed one, traced it to Flock’s website, and decided the public deserved a map of what was watching it. He built one on OpenStreetMap, the open-source mapping database, so contributors anywhere can plot license plate readers, note the brand, and record the direction each camera faces. The directional data matters: it revealed, for example, that Huntsville’s downtown cameras all point outward, ringing the core to log every car entering it. After 404 Media covered the project in November 2024, it grew from a few dozen mapped cameras to thousands within weeks, and it now documents tens of thousands of readers across the United States and worldwide, covering Flock, Motorola, Avigilon, and other vendors. The architecture is itself a governance lesson. Because the data lives in OpenStreetMap rather than on DeFlock’s servers, no lawsuit or takedown can erase it; the community reverts vandalism the way Wikipedia does, and anyone can rebuild the site from the same data. That resilience has already been tested. Less than a year after launch, Flock sent DeFlock a cease and desist built on trademark claims the project calls bogus, and Flock’s CEO publicly likened the mapmakers to a terrorist operation, in remarks captured on video. Neither move removed a single pin. Freeman’s stated goals are modest and civic: awareness first, signage and oversight second, removal where communities choose it, and eventually navigation that lets a driver route around known cameras. A volunteer with a map has produced more public knowledge about this network than every transparency portal the vendor operates, which tells you who is actually doing the accountability work.
The third tool is law with teeth, and the operative word is teeth. Virginia’s 21-day retention statute demonstrably shaped the Norfolk litigation. Illinois’s sharing ban caught CBP, but only because the Secretary of State chose to audit, and it imposed no penalty when violated; the next version of that law, anywhere, should carry per-violation fines, a private right of action, and contract debarment for vendors who breach it. Washington jurisdictions that restricted access to in-state agencies escaped the federal back door entirely, a controlled experiment in the value of default-closed settings. Warrant requirements for retrospective database searches, mandatory public error reporting on the Oak Park model, automated flags on repeat searches of a single plate, and a flat statutory ban on selling or ingesting breach-derived data would close most of the pathways this article documents. None of that requires new constitutional doctrine. It requires legislators who read audit logs, and voters who ask whether they have.
And the courts may yet do their part. Chatrie’s holding, that government access to a private company’s location database is a search regardless of how small a slice is taken, is the first Supreme Court reasoning that maps cleanly onto what Flock built. Schmidt v. City of Norfolk will test it at the Fourth Circuit. The San Jose case will test state constitutions. However those cases land, communities should notice that every judicial victory Flock has won so far rests on the network being not-quite-total yet, and the company’s growth plan is to remove the qualifier.
Security Over Liberty Gets You Neither
Strip away the technology and the question underneath is old. Benjamin Franklin’s much-abused line about trading liberty for safety gets misquoted in both directions, but the paraphrase that opened this assignment is the durable one: a society that values security over liberty ends up with neither. Flock is the cleanest modern test of that proposition available, because the ledger can now be read on both sides.
On the liberty side, the losses are documented above and need only be totaled. Americans in 6,000 communities are logged, timestamped, and fingerprinted by vehicle roughly 20 billion times a month without suspicion, warrant, or meaningful consent. Their movements have been queried nearly 4 million times in a year in a single city. Their data has been searched for attending protests, for belonging to an ethnic group described in slurs, for driving while an officer’s ex, for writing a newspaper column, for having an abortion. The sharing controls failed. The audit logs recorded abuse without preventing it. The state laws lacked penalties. The vendor’s public statements had to be corrected repeatedly, sometimes by the vendor itself, and it coached its customers toward vagueness in the logs meant to catch abuse. Whatever liberty means, the freedom to travel without generating a police-searchable dossier was part of it, and it is gone wherever these poles stand.
Now the security side, which is the side the whole trade was supposed to purchase. The Sherwood couple at gunpoint with their infant in the back seat were not made secure. Joel Feder, boxed in by 4 squad cars over a typo, was not made secure. Chrisanna Elser, criminally charged by camera and acquitted by her own doorbell footage, was not made secure. The woman on the Florida forest trail whose evening routine sat exposed on the open internet was not made secure. The 21-plus stalking victims tracked by the officers sworn to protect them were precisely dis-secured by the system, as was the Texas woman whose abortion triggered an 83,000-camera manhunt while her actual assailant held the phone that reported her. Even the aggregate safety dividend, the one thing that could justify some of this, remains unproven by any independent measure after 8 years, while the only town publishing error data shows a third of camera-triggered stops hitting innocent people. The security was supposed to be the compensation. The compensation check has not cleared, and the account it was drawn on belongs to the people it was promised to.
This is why the Franklin formulation is not a slogan but a mechanism. Surveillance systems do not merely cost liberty; they convert liberty into new categories of insecurity. Every database of everyone becomes a menu for the worst person with a login, and there is always a worst person with a login: the stalker lieutenant, the borrowed-password DEA agent, the sheriff with the grand jury problem, the misconfigured test unit, the checkbox nobody read. A town without Flock cameras cannot have its residents’ movements leaked, stalked, subpoenaed, or federally back-doored, for the same reason a safe that does not exist cannot be cracked. The most secure data is the data never collected. That is not privacy absolutism. It is engineering.
None of this requires believing that the police chiefs of America woke up wanting a panopticon. Most bought a product that promised to help, from a salesman with a recovered Kia slide, under pressure from residents frightened by a car break-in on Nextdoor. The failure was not malicious. It was the oldest institutional failure there is: officials accepting a power because it was offered, without asking who would hold it after them, and citizens outsourcing their safety to a subscription because outsourcing is easier than knowing the neighbors. The correction is equally unexotic, and it is already underway in Evanston and Eugene and Weston and El Cerrito: communities deciding, in public meetings, with the audit logs on the table, that they would rather solve somewhat fewer stolen-car cases than live on camera, and that the safety worth having is the kind neighbors provide each other with their eyes open, their records public, and their government asking permission. A camera never asked permission. That was the whole product. It should be the whole objection.
Sources and Further Reading
Institute for Justice, “Dozens of Innocent Motorists Have Been Pulled Over, Detained at Gunpoint, or Jailed Due to AI License Plate Camera Errors” (July 2026) and “Police Have Reportedly Used License Plate Readers to Stalk Romantic Interests at Least 21 Times” (June 2026); Schmidt v. City of Norfolk case materials
404 Media reporting by Jason Koebler and Joseph Cox: ICE lookups via Flock (May 2025); Texas abortion search (May 2025); CBP access to 80,000+ cameras (Aug. 2025); Nova and breached data (May 2025); death investigation court records (Oct. 2025); No Kings protest searches (Nov. 2025)
Electronic Frontier Foundation: “She Got an Abortion. So a Texas Cop Used 83,000 Cameras to Track Her Down” (May 2025); “How Cops Are Using Flock Safety’s ALPR Network to Surveil Protesters and Activists” (Nov. 2025); “Flock Safety and Texas Sheriff Claimed License Plate Search Was for a Missing Person” (Oct. 2025); 2025 Year in Review; San Jose lawsuit filings with ACLU of Northern California
University of Washington Center for Human Rights, “Leaving the Door Wide Open” (Oct. 2025)
Illinois Secretary of State audit announcements (Aug.–Sept. 2025); Capitol News Illinois; Wednesday Journal (Oak Park); Unraveled
Courthouse News Service, Schmidt v. City of Norfolk ruling coverage (Jan. 2026); EPIC case analysis (Feb. 2025); Texas Bar Journal on United States v. Martin; coverage of Chatrie v. United States (June 2026) and the Schmidt appeal, No. 26-1227
Forbes, “Flock Installed AI Cameras in This Small City and Claimed Crime Went Down. It Went Up.” (Feb. 2024)
NPR, “Why Some Cities Are Canceling Flock Contracts” (Feb. 2026); KCUR/KMUW reporting on Kansas and Missouri (June 2026); ACLU “Get the Flock Out” campaign tracker
The Drive (Joel Feder first-person account) and follow-on coverage (June–July 2026); InvestigateTV on the Elser case (May 2026)
WFLX/WPTV on exposed Condor camera feeds (Jan. 2026); Straight Arrow News on Ring-Flock risks and camera vandalism investigations; IPVM, “Flock-Powered Police Chiefs Stalking Women” (June 2026)
NBC News, Engadget, Fortune, and The Hill on the Ring partnership and cancellation (Oct. 2025–Feb. 2026)
DeFlock (deflock.me / deflock.org), open-source ALPR map founded by Will Freeman, Oct. 2024; 404 Media, “The Open Source Project DeFlock Is Mapping License Plate Surveillance Cameras All Over the World” (Nov. 2024); DeFlock blog, “Why We Use OpenStreetMap” (Apr. 2025)
Congressional letter of Reps. Garcia and Krishnamoorthi to Flock (Aug. 6, 2025); military.com and WAVY on the Sovern case; Boulder Reporting Lab; RANGE Media; 9News Denver; Flock Safety’s own blog posts, white papers, and transparency materials
A note on sourcing: several late-breaking items above, including the June 2026 Supreme Court decision in Chatrie, the December 2025 Nova code analysis, and the April 2026 Dunwoody community center report, rest on the coverage cited and should be re-verified against primary documents before republication. Standard practice for a piece of this kind also applies: build a source file pairing every figure, quotation, court holding, and audit finding with the underlying record and an archived copy, and request comment from Flock Safety before publication, appending any response received.