The Town That Watches Everyone

CORRECTION AND UPDATE — AUGUST 21, 2026

An earlier version of this article reported that the Order Form — Exhibit A to Glenrock's contract with Flock Safety, the document that sets the term, the price, and the data retention period — appeared to be missing from the town's records production, and that pages GLENROCK_00019 through 00022 contained no readable content.

That was wrong. The Order Form is in the production. Those pages are image-only scans with no searchable text layer, so every text extraction I ran returned them blank. Rendered as images, they are fully legible.

I have now read them, and this version of the article is rewritten around what they say. The Order Form answers several questions the earlier version had to leave open, and it raises one the earlier version did not know to ask: Glenrock's contract does not renew for one year. It renews for twenty-four months.

The town produced this document. I failed to read it. The correction is mine.

Everything else in the original article stands. Section-by-section changes are listed at the end of this piece.

At 9:39 on the morning of December 8, 2025, Glenrock Police Chief Colter Felton ran a search against the town's Flock Safety license plate camera system. The reason he logged was "Hit and Run." He was looking for a white Ram pickup, in a window running from 6:00 to 10:00 that morning. That is exactly what this technology is supposed to be for, and nobody should begrudge him the search.

Two hours later, at 11:48 a.m., he started searching again. Over the next four hours and thirty-five minutes he ran forty-eight more queries. Not one of them carried a case number. Every single one was logged with the same reason: "Test."

Six of those "Test" searches filtered for a Ram. The first four used a time window running from 6:00 a.m. on December 4 to 9:30 a.m. on December 8 — a four-day span ending at the hit-and-run window. The last two narrowed to that morning alone, 6:00 to 9:30. In between and around them he cycled through a second vehicle description: a Chevrolet SUV, black, then unfiltered, then just "Wyoming" plates with no vehicle description at all, across windows from one hour to four days.

I cannot tell you with certainty what Chief Felton was doing that afternoon. Neither can the Town of Glenrock, and neither can you — and that is the actual story. The audit log is the one mechanism that is supposed to make a system like this accountable to the people it watches. In Glenrock, by the department's own records, it does not work.

HOW I GOT THESE RECORDS

I did not file the request that produced these documents. A source did — a public records request under the Wyoming Public Records Act, sent to the Town of Glenrock on July 17, 2026, asking for contracts, policies, search logs, and a sample of what the cameras actually capture. The town retained outside counsel in Casper — attorney Amy M. Iberlin — to respond. On August 19, 2026, that response came back: the Flock contract, the invoice, the department's written camera policy, and twenty months of audit logs covering every search an officer ran from October 2024 through June 2026.

My source passed the entire production to me. I have read all of it, and I have counted it — not by eye, but by extracting every row of the audit log into a spreadsheet and running the numbers. What follows is drawn from those documents, cited so you can check me, plus reporting from around the country on the same company's network.

The request is still open. The town's attorney wrote on August 19 that she had asked for all emails that in any way relate to "flock" — a batch she described as voluminous — and that reviewing the remainder would take about three to five more days. If she is on schedule, the next tranche of documents lands before the town's public meeting on Tuesday the 25th. Somebody should ask for it in the room, and the Council should not treat a meeting held that soon as the last word on a production that isn't finished.

Every document I cite here is posted publicly, in full, at this link:

The complete Glenrock Flock production — contract, invoice, policy, and all twenty months of audit logs

I'm not asking anyone to take my word for any of this. The Bates numbers in this piece — GLENROCK_00001 through GLENROCK_00074 — are stamped on the pages themselves. Look up any claim I make. If I've read something wrong, I want to know before Tuesday, not after.

I mean that literally, and this version of the article is the proof. The Order Form correction at the top of this piece exists because I said a document was missing when it was sitting in the file I had published.

THE MEETING IS AUGUST 25

The Town of Glenrock has scheduled a Flock Safety Informational Meeting for Tuesday, August 25, from 5:30 to 6:30 p.m. at Town Hall, listed on the town's own calendar as a public question-and-answer session on the camera system. The contact on the listing is Town Clerk Tammy Taylor.

A resident who posted about the meeting on August 12 said the 5:30 start was a compromise: the town couldn't fit it in before the election season — Wyoming's primary ran August 18 — and couldn't push it later in the evening. If you work in Casper or Douglas, a one-hour window at 5:30 on a Tuesday is a narrow door for a decision this size.

Mayor Bruce Roumell and the Town Council — Judi Colling, Roy Kincaid, Russ Dalgarn, and Margaret Nunn — hold the contract. Chief Felton signed it and administers the system. The town has not published a list of who will be at the table, so call Town Hall before you go if you want to know who you'll be asking.

I have not yet put these findings to Chief Felton or the Council for response. I'm publishing them before the meeting rather than after, because the point is for residents to walk in already knowing what to ask. If the department has answers, the meeting is the place to give them, and I will report them.

WHAT GLENROCK ACTUALLY SIGNED

The contract is a Master Services Agreement with Flock Group, Inc., an Atlanta company that dominates the license plate reader business. Chief Felton executed the Order Form on July 24, 2024, attesting that he had read and agreed to the attached Master Services Agreement (GLENROCK_00023). Flock's signatory was its general counsel, Mark Smith, the same day.

The Order Form — what the town actually bought

This is the document I previously reported as missing. Here is what Exhibit A says (GLENROCK_00020–00022):

Customer: WY – Glenrock PD, 219 S 3rd St, Glenrock, WY 82637

Accounts payable contact: cfelton@glenrock.org

Initial Term: 24 months

Renewal Term: 24 months

Retention Period: 30 days

Payment terms: Net 30

Billing: Annual plan, first year invoiced at signing

Hardware: Flock Safety Falcon ® × 5; FlockOS Essentials × 1

Annual recurring: $15,000.00

One-time, Standard Implementation Fee: $650.00 × 3 = $1,950.00

One-time, MASH-Tested Pole Implementation Fee (non-coastal): $1,250.00 × 2 = $2,500.00

Year 1 at signing: $19,450.00

Contract total: $34,450.00

Several things fall out of that immediately.

The renewal is two years, not one. The Master Services Agreement says the contract automatically renews for successive terms of "the greater of one year or the length set forth on the Order Form" unless either party gives thirty days' notice before the end of the current term (§ 7.1, GLENROCK_00010). The Order Form sets that length at twenty-four months. If Glenrock misses the notice window, it is committed for another two years, not one. This is the single most consequential fact in the production, and it was invisible until these pages were read.

Retention really is thirty days. The MSA's thirty-day purge applies "except as otherwise stated on the Order Form" (§ 1.14, GLENROCK_00003). The Order Form states thirty days. There is no secret longer retention period hiding in Exhibit A. That is a point in the town's favor and I'll say so plainly.

It also sharpens a question raised later in this piece. On November 17, 2025, an officer ran a search across a 108-day window (GLENROCK_00056). Against a database the contract says holds thirty days, roughly seventy-eight of those days should not exist. That is now unambiguous rather than speculative.

The fee structure tells you something about the poles. Three cameras were billed at the standard implementation fee; two at the "MASH-Tested Pole" fee. MASH is the federal crash-test standard for roadside hardware. The reasonable reading is that three cameras went onto existing structures and two required new crash-rated poles — consistent with the Wyoming Department of Transportation right-of-way permit Glenrock obtained. It is not a location list, but it is more than the town has otherwise disclosed.

And the real number is $34,450. Not "about $15,000 a year." The contract total across the initial twenty-four-month term is $34,450, front-loaded with $19,450 due at signing.

What FlockOS actually includes

Page GLENROCK_00022 is the product description, and it is the most revealing page in the contract. It lists what Glenrock's subscription includes:

Nationwide Network (License Plate Lookup Only) — described as letting agencies leverage a national system carrying ten billion additional plate reads per month.

State Network (License Plate Lookup Only).

Law Enforcement Network Access — the ability to request direct access to evidence detection devices held by agencies outside Glenrock's jurisdiction.

Community Network Access — the ability to request direct access to feeds from privately owned Flock cameras located in neighborhoods, schools, and businesses.

Real-Time NCIC Alerts on Flock ALPR Cameras — automated alerts when a vehicle is detected that matches the FBI's National Crime Information Center or NCMEC databases.

Unlimited Custom Hot Lists — the ability to add a plate to a custom list and be alerted when it passes a camera.

Vehicle Fingerprint search (type, make, color, and features like bumper stickers, decals, roof racks), time-and-location search, unlimited users, and an analytics module.

Three of those deserve a resident's attention.

"Unlimited Custom Hot Lists" is a purchased feature. Later in this piece I describe an alert logged as "Sex Offender 133195" — a category Glenrock's own written policy never contemplated. The Order Form explains how such a category can exist: the product permits unlimited custom lists, defined by whoever holds the login, with no policy constraint written into the contract.

"Community Network Access" reaches private cameras. Residents who assume this system is limited to five town-owned poles should read that line. The subscription includes the ability to request feeds from privately owned Flock cameras at businesses, schools, and neighborhoods.

The nationwide tier is plate-lookup only. That qualifier matters and cuts in the department's favor — a nationwide query is supposed to require a specific plate rather than a vehicle description. It also raises a precise question. Several Glenrock searches reached thousands of networks (see below). If the nationwide tier requires a plate, then a specific plate was entered on those queries — and the plate column is redacted, so residents cannot see what it was. That is a fair question for Tuesday, not an accusation.

Terms that deserve translation out of contract English

Glenrock does not own the cameras. Flock does, and the town isn't permitted to remove, reposition, or adjust one; any later change to the deployment plan incurs a fee under Flock's published reinstall schedule (§§ 10.1, 10.3, GLENROCK_00013–14). The town is a tenant of the surveillance infrastructure on its own streets.

Flock gets a permanent license to work with your driving data. The town keeps title to the footage, but grants Flock an "irrevocable, worldwide" license to use it to provide the services (§ 4.1, GLENROCK_00006). Separately, Flock has the right to anonymize that data and use it "including the training of machine learning algorithms," under a license the contract calls "perpetual" (§ 4.3, GLENROCK_00007).

The contract says that anonymization is irreversible. Maybe it is. The point isn't that Flock is lying — it's that no one in Wyoming can check. Glenrock has no audit right over Flock's anonymization, no ability to inspect the pipeline, and no way to revoke the license if the standard turns out to be weaker than advertised. The town signed away, in perpetuity, a thing it cannot verify and cannot take back.

The contract is broader than the policy. The agreement Glenrock signed describes a platform "capable of capturing audio, video, image, and recording data" and defines the services to include "audio detection" (Recitals, GLENROCK_00001; § 1.10, GLENROCK_00002). Glenrock's own policy assures residents that query results "do not contain audio/video recordings" (§ E.4.b, GLENROCK_00038).

The Order Form settles the practical question in the town's favor: what Glenrock bought is five Falcon LPR units and FlockOS Essentials. No audio product appears on the invoice or in the feature list. Today, the town is not buying audio detection. But the legal authority sits in the signed agreement, and § 1.10's definition of the services already includes it. Adding it would require no new contract and no new vote — only a change order.

Thirty-day deletion still has a hole in it. The Order Form's thirty-day retention governs the Flock cloud. The contract separately states that the customer "shall be responsible for extracting, downloading and archiving Footage from the Flock Services on its own storage devices" (GLENROCK_00001). Anything an officer downloads into a case file lives outside the thirty-day clock. Neither the policy nor the Order Form says anything about local copies.

If something goes wrong, Glenrock is on its own. Flock's liability is capped at twelve months of fees and expressly excludes "INCOMPLETE, CORRUPT, OR INACCURATE DATA" (§ 9.1, GLENROCK_00012) — a real exposure, given that plate misreads have produced wrongful stops and arrests elsewhere. Flock also reserves the right to hand footage to "law enforcement authorities, government officials, and/or third parties" if it has "a good faith belief" that doing so is reasonably necessary (§ 5.3, GLENROCK_00008–09).

Flock carries substantial insurance; the town's exposure is separate. Exhibit B (GLENROCK_00024–00025) requires Flock to maintain $1 million general liability, $10 million umbrella, $5 million professional liability, $1 million auto, and $5 million cyber liability. Worth knowing. Worth also knowing that § 9.1's liability cap and data-accuracy exclusion sit in the same contract, and that insurance protecting Flock is not the same as recourse for a resident wrongly stopped.

THE AUDIT LOG IS THE STORY

Glenrock's policy builds its entire accountability structure on one thing: the audit trail. The Chief, as LPR administrator, must ensure maintenance of an audit trail of every query, "which will be kept permanently," including "the name of the member," "the date and time of query or access," "the specific information queried or accessed," and "the authorized law enforcement justification for query or access, including a relevant case number" (§ D.1.f, GLENROCK_00036).

Here is what twenty months of that audit trail actually contains.

375 searches. One case number.

The log holds 375 searches (GLENROCK_00040–00074), run by five officers: Chief Felton (161), Officer Rachel Brurud (134), Officer Jesse Richard (49), Officer Sabryna Schaffer (19), and Officer Jeff Tesdall (12).

Five of the 375 carry a case number. All five are the same case — G25050108, a May 2025 matter logged as "government threats," run by Officer Tesdall (GLENROCK_00049).

Every other search in the production — 370 queries into where residents' vehicles have been — has an empty case number field. The policy requires one. This is not a close call or a matter of interpretation.

More than one in four searches say "Test" or "Training"

Ninety-eight entries — 26.1% of the log — are logged with a reason of "Test," "test," "Training," or "training." Eighty-six say some form of "Test"; twelve say some form of "Training." Chief Felton ran 92 of them. Officer Brurud ran the other six.

To be fair to the department: the policy does authorize training queries. Section E.1.f permits searches "For the proper training of new users or for the purpose of maintenance/ calibration" (GLENROCK_00037). Training on a live system is legitimate, and a small department has to train somehow.

But look at the distribution across the Chief's own 92.

[CHART: Chief Felton's "Test"/"Training" queries by month, Oct 2024 – June 2026]

The October and November 2024 entries are entirely plausible for a new user learning a new system. The 48 in a single afternoon in December 2025 — fourteen months in, by the system's administrator — are something else. So are the ten in May 2026 and five in June 2026, twenty months in.

E.1.f authorizes training for new users. Section H.1 requires that training happen before access is authorized (GLENROCK_00039). Neither provision obviously covers the person who signed the contract running four dozen unlabeled queries in one afternoon a year and two months later. And whatever E.1.f permits, D.1.f still requires a justification and a case number on every query. "Test" is a label, not a justification.

The department kept picking a real crime — and then typing "Test"

Between December 17, 2025 and January 8, 2026, the log format changed. Earlier entries have a single free-text reason. Starting January 8, 2026, entries carry a structured crime category plus a note: "Motor Vehicle Theft/Stolen - Stolen vehicle," "Drugs/Narcotics - Drug Investigation," "Assault/Battery Offenses - Investion" [sic]. Flock appears to have begun forcing officers to select an offense type from a dropdown.

Glenrock's response to that guardrail is visible in the log:

"Gambling - Test" — ten searches on May 27, 2026 (GLENROCK_00070)

"Trespass - Test" — five searches on June 2, 2026 (GLENROCK_00074)

An officer selected a real criminal offense from the dropdown and then typed "Test" as the reason. Both things cannot be true. Either these were genuine tests, in which case the department fed false offense data into the system that Flock and the town publicly report from — or these were real investigations into gambling and trespass, in which case the stated reason is false. There is no third reading in which the log is accurate.

For what it's worth, the May 27 "Gambling" sequence does look like feature-testing. Over twenty-five seconds, from 14:24:46 to 14:25:11, the body-type filter walks methodically down a list — Sedan, Coupe, Convertible, Hatchback, Wagon — then drops one at a time until only "Coupe" remains, then nothing. That is what someone poking at a filter looks like. Which makes it all the stranger that "Gambling" was attached to it.

And note the sequence on June 2. The properly-logged searches came first: Chief Felton ran seven queries between 10:19 and 10:53 that morning under "Disorderly Conduct/Disturbance - Identification." The five "Trespass - Test" queries came that afternoon, between 3:01 and 3:13. The department demonstrated it knew how to fill the field correctly, and then, four hours later, didn't.

Four months with no searches at all

Four pages of the produced audit log are screenshots of Flock's Organization Audit interface returning "No data found for report": March 2025 (GLENROCK_00047), July 2025 (GLENROCK_00051), August 2025 (GLENROCK_00052), and October 2025 (GLENROCK_00054). Those months are genuinely empty.

There's a fifth gap worth asking about. The records request asked for logs "since installation." The Order Form was executed July 24, 2024, and § 7.1 provides that the term commences upon first installation of the hardware. The first logged search is October 15, 2024 — nearly three months later. Either the cameras weren't operational until October, or roughly a quarter of a year of logs is missing. The town should say which, and the answer also fixes the renewal date.

The timeframes keep stretching

The most consistent pattern in the log is officers starting with a tight window around an incident and then widening it, sometimes dramatically, within seconds.

A larceny, September 22, 2025 (Officer Brurud, GLENROCK_00053). At 4:46:17 p.m. she began with a same-day window of about seventeen hours, looking for a white Honda SUV. Seventeen seconds later, a two-day window. Thirteen seconds after that, a twenty-one-day window running back to September 1. Thirty seconds, start to finish: seventeen hours to three weeks — for a theft. She kept working the twenty-one-day window for another four minutes, adding silver to the color filter.

A disorderly conduct call, June 2, 2026 (Chief Felton, GLENROCK_00074). This one is logged "Disorderly Conduct/Disturbance - Identification" — the purpose was to identify a person, not to locate a known suspect vehicle. He started at 10:19 a.m. with a 7.5-hour window on May 30, hunting a green coupe, and tried several body-type combinations. By 10:53 a.m. he had dropped the color and body filters entirely and widened the window to sixteen days, back to May 17 — searching Wyoming plates generally.

The December 8 hit-and-run sequence described at the top of this piece: a four-day window, then a narrowing to the incident morning, all of it labeled "Test."

And one window that should not be possible. On November 17, 2025, Officer Brurud ran a drug-investigation lookup with a timeframe running from August 1 to November 17 — 108 days (GLENROCK_00056). The Order Form sets retention at thirty days. Roughly seventy-eight of those days should be empty. Why would an officer search a window three and a half times longer than the data can exist? Either the interface simply accepts impossible queries, or something is retained longer than the contract provides. That is a direct, answerable question, and it deserves a direct answer on the 25th.

Searches that leave Wyoming — by a lot

The log records how many separate camera networks each search touched. Most Glenrock queries hit one — the town's own. Sixty do not.

November 15, 2025, 7:59 p.m.: Officer Brurud ran four consecutive drug-investigation searches reaching 6,281 separate camera networks (GLENROCK_00056). The log timestamps these in UTC as November 16; Wyoming was seven hours behind.

March 25, 2026, 9:26 a.m.: Chief Felton ran a stolen-vehicle lookup reaching 5,794 networks (GLENROCK_00067).

May 7, 2025: eleven searches at 48 networks, including two of the five case-numbered "government threats" queries (GLENROCK_00049).

Dozens of routine drug, theft, warrant, and assault queries reaching 22, 35, 37, 38, 41, 42, 43, 44, 45, 46, or 47 networks (GLENROCK_00055–59, GLENROCK_00064, GLENROCK_00067–68, GLENROCK_00072).

One of those is worth pausing on. On December 6, 2025, Officer Brurud ran a search logged only as "Test" that reached 46 networks (GLENROCK_00064). Whatever "Test" means, on that occasion it reached forty-six other agencies' cameras.

The Order Form now tells us this capability was purchased, not stumbled into: "Nationwide Network (License Plate Lookup Only)" is a listed feature of Glenrock's FlockOS subscription (GLENROCK_00022). The Council bought national reach. Whether anyone on the Council knew that is a separate question.

For scale: when the Electronic Frontier Foundation reviewed Flock's national audit data, the case that drew the most alarm was a Texas deputy who searched 6,809 networks and 83,345 cameras in an investigation that began after a woman was reported to have had an abortion. (The sheriff's office and Flock publicly characterized it as a welfare check for a missing person; EFF obtained documents showing deputies had discussed criminal charges.) Glenrock's 6,281-network drug lookup is the same order of magnitude. A town of 2,400 people is running searches at national scale.

One entry deserves care, because it looks worse than it is. On November 27, 2024, a felony warrant search lists a long roster of jurisdictions — 49 states (every state but Wyoming), plus Washington D.C., Canada, and Mexico. That roster sits in the Filters column, not the networks column. It is a license plate jurisdiction filter: show me any plate from anywhere but here. The Total Networks Searched field for that search reads 1. It was a search of Glenrock's own cameras for out-of-state plates, not a continental dragnet.

The 6,281 and 5,794 figures are in the networks column. Those are real.

A hot list category the policy never contemplated

On May 5, 2025, at 2:50 p.m., the log records a search generated from an alert: "image search from alerts page associated with Alert: Sex Offender 133195," run by Chief Felton against a 29-day window (GLENROCK_00049).

Glenrock's policy defines a Hot List as stolen vehicles, stolen plates, AMBER and SILVER alerts, plates "associated with specific individuals, such as wanted individuals or missing individuals (e.g., wanted for homicide, rape, robbery, child abduction)," or terrorist watch lists (§ B.3, GLENROCK_00035). Every enumerated example involves someone actively wanted or missing.

A person on a public registry who has completed their sentence is not wanted for anything. If registry status generates an automatic alert every time that person drives past a camera, Glenrock has extended a "hot list" designed for active fugitives into permanent, automated monitoring of someone not accused of any current crime — a category its own policy never described. Whatever you think of registries, that is a decision the Council should have made knowingly, not one that arrived preloaded in a vendor's default settings.

The Order Form shows how it could arrive preloaded. Glenrock's subscription includes "Unlimited Custom Hot Lists" — the ability to add any plate to any list and receive an alert when it passes a camera (GLENROCK_00022). The product imposes no limit. The policy does. Nothing in the record shows the policy being enforced against the product.

The log shows one such alert-derived search. It does not show whether the alert was a one-off or a standing subscription. That's a question for Tuesday.

WHAT THE TOWN DID NOT PRODUCE

Sometimes the gaps matter more than the documents — but it's worth being precise about which gaps are real. This section is where my earlier version got it wrong, so I'm going to be careful.

The Order Form: produced, and I missed it. Reported previously as missing. It is at GLENROCK_00020, with the billing schedule at 00021 and the product description at 00022, behind a Flock cover page at 00019. All four are image-only scans with no text layer, which is why every search of the file came back empty. That is a real accessibility problem in a public records production — a document nobody can search is a document most people won't find — but it is not a withholding, and I should not have said it was.

How many cameras: answered, and easy to miss. The production discloses the count twice. Invoice INV-76999 (GLENROCK_00031) lists five LPR units at $3,000 apiece. The Order Form (GLENROCK_00020) lists "Flock Safety Falcon ®, quantity 5." A resident who read the policy and the entire audit log would find it in neither.

Where they are: still not answered by the town. The records request specifically asked for capture data including "camera location," for cameras "in or near both Glenrock and Rolling Hills." The response to that request directed the requester to Flock's public transparency portal instead. No list of intersections, no map, no addresses.

Here's what makes that non-answer strange rather than sinister: the locations aren't secret, and the town knows it. Chief Felton told Cowboy State Daily in July that his agency has cameras at every entrance to town, and that they're strategically placed so the department knows if a vehicle comes into or out of Glenrock. That same paper obtained Wyoming Department of Transportation permit records showing Glenrock, through Felton, secured a permit to mount Flock readers on highway rights of way north of the Rolling Hills area. Anyone can also look up crowdsourced pins on DeFlock, the open-source ALPR map founded in October 2024 by Alabama software engineer Will Freeman, which now lists roughly 129,500 readers nationwide.

So residents can assemble an approximate answer from a newspaper, a state highway agency, and volunteers on the internet. What they cannot get is the authoritative one, from the government that bought the cameras, in response to a lawful written request. There is no security justification for that. The only thing the refusal accomplishes is making the official record less accurate than a hobbyist map.

If the town's position is that camera locations are sensitive, it should say so and explain why. If the position is that they're public, it should simply publish the list: five cameras, these five poles. The Order Form's fee structure suggests two of them sit on new crash-rated poles and three on existing structures. The town can confirm that in one sentence.

And DeFlock is not a substitute. Cowboy State Daily checked it against three Wyoming towns in July: right about Lusk, wrong about both Cody and Gillette, where local agencies said they use no readers at all and WYDOT suggested the mapped devices were traffic-signal equipment. Crowdsourced is not authoritative. That is an argument for the town publishing its own list, not against it.

A plate slipped through the redactions. The town's attorney wrote that "Personal identifying information, such as License Plate Nos., have been redacted." On GLENROCK_00065, one row carries what appears to be a plate — not in the License Plate column, which is properly blacked out elsewhere on the page, but typed into the Filters field as free text. The redaction pass caught the structured column and missed the free-text one. I'm not republishing it. But if a careful law firm reviewing a 74-page production can leak a plate, it is worth asking what happens to this data at scale, in a system where thousands of agencies have access.

And the department has closed the obvious accountability door. Glenrock's policy states that under Wyoming law, "the Department does not create records to fulfill public records requests and no query will be performed (and document created) in response to a public record request" (§ F.3.b, GLENROCK_00039). Read that again. If you want to know whether your car has been searched, the department's stated position is that it will not look. The only public window is Flock's own transparency page — thirty days, rolling, published by the vendor being scrutinized. It appears nowhere in the contract and nowhere in the policy; the town has no enforceable right to it, and Flock can change or remove it at will.

"LITTLE DRAMA" — AND HOW GLENROCK ACTUALLY FOUND OUT

On July 3, 2026, Cowboy State Daily published a piece on this town's cameras under a headline announcing that, unlike Cheyenne, Jackson, and the Wind River Indian Reservation, Glenrock's Flock cameras "bring little drama." Reporter Clair McFarland had contacted Chief Felton after WYDOT permit records surfaced Glenrock's installation.

Felton told the paper the town had seen almost no pushback: "We really didn't get a ton of, I guess, people showing up to talk about it, or with any sort of issue." He said one person came to a meeting held in Rolling Hills and had some discussion. He credited community trust in local law enforcement for the calm, and separately theorized that much of the national outrage traces to misinformation about what the cameras do.

I don't doubt any of that is accurately reported. Here is the problem with it.

Silence was treated as consent. But in a town of about 2,400 people, one attendee at one meeting is not a community that considered the question and approved. The clearest evidence of that is in the same article: State Rep. Kevin Campbell, who represents Glenrock in the Wyoming Legislature, told the paper he first learned the cameras existed about a year ago — roughly a year after they went up. If the town's own state legislator didn't know, the inference that residents knew and were fine with it does not survive contact with the record.

The written policy governing these cameras was not posted on the town's website until August 2026. The URL the town's attorney supplied — glenrock.org/wp-content/uploads/2026/08/Glenrock-PD-Flock-LPR-Policy.pdf — carries its own upload date in the path. For roughly two years, the rules governing a surveillance system on Glenrock's streets were not published anywhere a resident would look.

And consider the sequence of the last two months. The Cowboy State Daily piece ran July 3. Twelve days later, the same outlet ran a follow-up examining DeFlock's map of license plate readers across Wyoming. The records request that produced these documents was filed July 17, two days after that. The town scheduled a public informational meeting for August 25.

From what I can tell talking to people here, that is the actual order of events for most of Glenrock: residents did not find out about these cameras from their town. They found out from a news article about how untroubled their town was, or from an online map built by a stranger in Alabama. The story written to document Glenrock's calm is a significant part of what ended it.

I'd rather be corrected than assume, and this is a claim that can be tested in the room. On Tuesday, ask for a show of hands: who in this room knew, before this summer, that these cameras existed, who ran them, what a "hot list" was, or that a Glenrock officer could query camera networks in other states? If most hands go up, I'm wrong and I'll say so in print.

None of this is a knock on the reporter who wrote it or on the chief who answered the question honestly. It is a caution about the specific inference: nobody complained and nobody objects are different sentences, and only one of them is evidence of anything. The fact that the town has now scheduled a meeting suggests somebody at Town Hall already understands the difference.

Rolling Hills

Glenrock's police department polices the neighboring town of Rolling Hills under contract, worth $73,745 a year in Glenrock's budget (GLENROCK_00027). Rolling Hills has roughly 440 residents, per the most recent census. The WYDOT permit Glenrock obtained covers highway right-of-way north of the Rolling Hills area. They are covered by this camera system. They are not represented on the Glenrock Town Council, they don't vote in Glenrock's elections, and they weren't party to the contract Chief Felton signed.

It is worth being precise about the one piece of public participation anybody can point to. Chief Felton told Cowboy State Daily that one person came to a meeting held in Rolling Hills and "had some discussion." That is the whole of it. Nothing in the record establishes where that person lived, what they said, or whether they objected. It has since been repeated around town as though a Rolling Hills resident showed up to a Glenrock council meeting and lodged a protest. The source does not say that.

The structural point doesn't need the anecdote: 440 people are policed by this system and have no vote on whether it renews — including on a renewal that now appears to run two years.

THE FOURTH AMENDMENT PROBLEM

Nobody disputes that a police officer can look at your license plate on a public street. That has always been true, and it is why defenders of this technology say there's no privacy interest at stake.

That argument works for one glance. It stops working when the glance is automated, permanent, indexed, and searchable across a national network.

The Supreme Court drew this line in Carpenter v. United States in 2018, holding that police need a warrant for historical cell-site location records because assembling a person's movements over time reveals "the privacies of life" — where you worship, who your doctor is, whose house you stayed at. The individual data points are innocuous; the compilation is not. Courts call it the mosaic theory, and it is precisely what a twenty-one-day larceny search or a sixteen-day disorderly-conduct search produces.

The Court pushed further this summer. In Chatrie v. United States, decided June 29, 2026, it held 6–3 that when police obtain a person's digital location history — there, a geofence request to Google for every phone near a bank robbery — that is a Fourth Amendment search. Writing for the majority, Justice Kagan said an individual has a reasonable expectation of privacy in records about his cellphone's location, and that police intrude on that interest when they demand the information, even for a limited time and even from a third party.

The holding is worth stating precisely, because a lot of coverage has overstated it. The Court held that a search occurred. It did not rule on whether the warrant in that case was valid; it sent the case back to the Fourth Circuit to decide whether the search was reasonable. What the ruling establishes is the threshold: this kind of data compilation is the sort of thing the Fourth Amendment governs at all. The warrant question follows, but it wasn't answered.

Michael Soyfer of the Institute for Justice — which has active federal lawsuits challenging warrantless ALPR use in Norfolk, Virginia and San Jose, California — read the decision as focused squarely on the mass surveillance character of these technologies. What worries him, he told reporters, is police being able to go back in time and reconstruct someone's movements despite having had no reason to suspect them when the data was collected.

That describes Glenrock's system precisely. And Glenrock's internal standard doesn't come close to a warrant. The policy permits a query on "reasonable suspicion" (§ E.1.c, GLENROCK_00037) — the standard for briefly stopping someone on the street, not for searching them. The department has borrowed a street-stop threshold and applied it to a retrospective location search of everyone who drove through town. And as the log shows, it is not consistently documenting even that lower bar: 370 of 375 searches have no case number at all.

Wyoming's own constitution may go further than the federal floor. Article 1, Section 4 protects against unreasonable searches and seizures in language independent of the Fourth Amendment, and a number of state supreme courts have read their own charters as more protective of location privacy than current federal doctrine. Whether Wyoming's courts would do the same for plate reader data is, as far as I can determine, untested. It is a question worth putting to a Wyoming civil liberties attorney, and I intend to.

THE FIRST AMENDMENT PROBLEM THE DEPARTMENT ALREADY WROTE DOWN

Glenrock's own policy prohibits using the cameras for:

"The purpose of prohibiting, infringing upon, or deterring activities protected by the First Amendment, such as freely practicing one's religion, freedom of speech and peaceful assembly, freedom of the press, and the right to petition the government for the redress of grievances." (§ F.2.c, GLENROCK_00039)

It separately bars queries based on someone's "religious, political, or social views or activities," their "participation in a particular noncriminal organization or lawful event," or their race, ethnicity, age, disability, gender, gender identity, sexual orientation, pregnancy status, or other protected classification (§ E.2.b, GLENROCK_00037).

Police departments do not write sentences like that about radar guns. They wrote it about this system because this system can do that, and because on this same national network, it has:

EFF obtained records covering more than 12 million searches by more than 3,900 agencies between December 2024 and October 2025. More than fifty federal, state, and local agencies ran hundreds of searches connected to protest activity — the 50501 protests in February, Hands Off in April, No Kings in June and October.

More than eighty agencies ran searches using slurs and dehumanizing language targeting Romani people, frequently with no crime specified at all.

The ACLU has documented Flock's repeated misstatements to city councils, including in Oshkosh, Wisconsin, where the council approved a contract and then revoked it the next day after learning the system could in fact produce a "heat map" of an individual vehicle's movements.

Federal immigration agents have accessed local Flock networks, in some cases without the local department's knowledge, prompting litigation in California and a push from Sen. Ron Wyden for enforceable protections against immigration- and abortion-related misuse.

None of that happened in Glenrock. I want to be clear about that, because it would be dishonest to imply otherwise. Glenrock's log is mostly small-town police work: a missing juvenile in late November 2025, felony warrants, stolen vehicles, DUIs, an animal cruelty investigation, a hit and run. Much of it is exactly what residents would want their police doing.

But note what's in Glenrock's log anyway. A case logged as "government threats" — the only case in twenty months that carries a case number — searched across forty-eight networks. A "Disorderly Conduct/Disturbance - Identification" search that widened to sixteen days of Wyoming plates. And on June 3, 2026, a search logged under the category "Disturbing Public Peace/Riot" (GLENROCK_00073).

Threatening a public official is a real crime and police should investigate it. But "government threats" is also the exact category where angry political speech gets reported as a threat. "Disorderly conduct" and "disturbing the peace" are the charges most often used at protests. "Identification" — using the camera network to figure out who someone is — is the capability that makes a crowd searchable. These are not accusations against Glenrock's officers. They are a description of how thin the line is, and of the fact that the only thing standing on it is the good judgment of whoever holds the login.

And here is where the First Amendment clause meets the audit trail. That prohibition has exactly one enforcement mechanism: the log. If the log carries no case number on 370 of 375 searches and labels one in four a "test," then no resident, no council member, and no judge can ever detect a query run against the cars parked outside a church, a union meeting, or a recall petition drive. An unauditable prohibition is a sentiment. It is not a rule.

WHICH BRINGS US TO THE PART NOBODY WANTS TO SAY OUT LOUD

Every defense of this system rests on trust. Chief Felton told Cowboy State Daily in July that he'd like to believe Glenrock's calm reflects the trust residents have in local law enforcement. He may well be right about that trust, and he may well have earned it.

That is not the question.

The question is what happens to this infrastructure when the people running it are not the people running it now.

Read Glenrock's policy for who checks the power. The Chief of Police is the LPR administrator (§ D.1, GLENROCK_00036). The Chief authorizes who gets access. The Chief conducts the audits that determine whether the system is being used properly. The Chief approves any sharing with outside agencies (§ F.3.a, GLENROCK_00039). There is no council review of searches. No civilian oversight board. No independent auditor. No judge. No warrant. The entire architecture of restraint is one office, checking itself, and publishing nothing.

That design works exactly as long as the person in that office is trustworthy. Chiefs retire. Councils turn over. Mayors change; this one's term expires this year. State and federal administrations change, and their enforcement priorities change with them — which is not a partisan observation, it is a description of every four-year cycle in American history. The people change. The cameras stay up. The database keeps filling. And the access, once granted, does not un-grant itself.

State Rep. Kevin Campbell, a Republican who represents this area, put it plainly to Cowboy State Daily: "Anytime we give something up we never get anything back; that's my biggest concern with them."

That is the whole argument, and it is not a partisan one. The Fourth Amendment was not written by men who distrusted government in the abstract. It was written by men who had lived under writs of assistance — general warrants that let British customs officers search any home, at any time, without naming what they were looking for. The colonists' objection was not that the particular officers were cruel. It was that the power was unbounded, and that a power unbounded will eventually find a wielder who uses it fully. They wrote a rule against the tool, not against the man holding it, because they understood the man would be replaced.

Glenrock has built, for $34,450 over two years, a tool whose legal authority already exceeds its current use — the contract covers audio the town says it doesn't collect; the subscription includes unlimited custom hot lists and access to privately owned cameras; the search reach already extends to six thousand networks the Council has never seen. It has connected that tool to a national network whose documented failures include tracking protesters, profiling an ethnic minority, and reaching into a woman's medical decisions. And it has secured the whole arrangement with an audit trail that — by the town's own records — contains one case number in twenty months and labels one search in four a "test."

If you believe the current chief would never misuse it, fine. Assume he wouldn't. The question on August 25 is not whether you trust Colter Felton. It's whether you'd hand these same capabilities, with these same non-existent checks, to someone whose name you don't know yet — for another two years.

DOES IT EVEN WORK? NOW WE CAN CHECK

Worth asking, given the tradeoff. And as of this week, the numbers exist.

Glenrock PD reports its offense data monthly to the Wyoming Division of Criminal Investigation, which publishes it by agency through its Crime Insight portal at crimestats.wyo.gov. Anyone can select "Glenrock PD" from the jurisdiction menu and pull it. Here is what it shows.

The cameras went live in October 2024. The first search in the town's own audit log is October 15, 2024. That makes 2025 the first full year of operation, and 2021 through 2023 the clean baseline.

[TABLE: Glenrock PD reported offenses, 2021–2025, Wyoming DCI Crime Insight]

(The portal displays 2025 case counts and the percentage change against 2024; the 2024 figures are derived from those percentages. Property-crime figures for 2021–2023 were not captured in the screenshots I worked from and should be pulled before anyone builds on them.)

So: in the first full year with license plate cameras at every entrance to town, reported property crime in Glenrock rose 71 percent, and larceny more than tripled.

Larceny is the exact crime class these cameras are sold to address. Flock's own marketing case — San Marino — was a burglary-and-theft story. In Glenrock, theft went from three reported offenses to ten.

Now the part where I refuse to write the obvious headline

The violent crime number went from 1 to 5. That is a 400% increase, it happened in the first full year of camera operation, and it would make a hell of a subheading.

I'm not going to write it that way, because it would be dishonest, and because it is precisely the move Flock made in San Marino — just pointed in the other direction.

Here's why. All five of Glenrock's 2025 violent offenses are sexual assaults. Aggravated assault went to zero. There were no homicides and no robberies. The five-year sexual assault line reads 1, 2, 1, 0, 5.

Sexual assault is overwhelmingly committed by someone the victim knows. It is not a crime that license plate readers detect, deter, or solve, and nobody at Flock claims otherwise. In a town of 2,400, a shift from zero reported cases to five is at least as likely to reflect a change in who felt able to report as a change in who offended — and if it is a reporting change, that's a good thing that has nothing to do with cameras. Attributing it to a surveillance system would be statistical malpractice.

The same caution applies, in fairness, to the property numbers. Seven offenses to twelve is five additional reported crimes in a town this size. It does not prove the cameras caused anything.

Which is the actual finding

The data cannot show that this system works. It can show that it hasn't.

Two years in, five cameras, $34,450 committed, 375 searches run — and the crime category the technology exists to address went up. Not dramatically, not causally, but unmistakably not down. There is no version of this table that a reasonable person reads as evidence the investment paid off.

And the burden here does not sit with residents. The town bought this system without publishing a justification, without a baseline, and without ever committing to a measure of success. Nobody at Town Hall has said what number would have counted as failure. That is not an oversight; it is what makes a program permanent. A thing that was never justified with evidence can never be un-justified by evidence.

One number would settle a great deal, and only the department has it. Property crime clearance in 2025 was 58.33 percent — larceny 60 percent, burglary 50 percent. Those aren't bad. But the portal doesn't publish pre-Flock clearance rates alongside them, and clearance is the only place a plate reader could plausibly show up in this data. So the question for Tuesday is not "did crime go down." It is:

Of the twelve property crimes Glenrock cleared or investigated in 2025, how many were cleared using Flock? And what was the department's clearance rate in 2022 and 2023, before the cameras?

The Chief can answer both from his own files. If the cameras are closing cases, that is a real argument and he should make it with numbers. If he can't produce the numbers, that is an answer too.

Meanwhile, the national evidence that these networks reduce crime is weaker than the sales deck suggests. Flock claimed a 70% crime reduction in San Marino, California. When Forbes checked the city's actual numbers three years after installation, residential burglaries had risen from 60 to 63 — a 5% increase — and overall Part 1 crime was flat: 230 offenses the year before installation, 231 three years after. The company had selected a favorable five-month window. San Marino's own police chief acknowledged the 70% figure wasn't accurate. Academics who reviewed the methodology were blunter; the Independent Institute's Jonathan Hofer called it "Exhibit A of lying with statistics."

A WEEK AGO, FLOCK CHANGED ITS OWN RULES

On August 13, 2026, Flock Safety announced a package of changes to its default settings, framed by CEO Garrett Langley as a response to nationwide backlash and to reporting on officers misusing the system. The relevant ones:

1. Default retention drops from 30 days to 7. Flock says more than 90% of searches that don't involve a complete plate happen within a week. Critically for Glenrock: existing customers keep their current retention periods unless they choose to opt in. Glenrock's Order Form sets thirty days (GLENROCK_00020), so unless the town affirmatively acts, thirty days is what it keeps. Agencies that adopt the seven-day window get the new "Evidence Mode" — which lets detectives preserve specific records for an active investigation — at no cost. Agencies that keep 30 days pay for it.

2. Case codes will be required for every law enforcement search by the end of 2026. Emergency searches can bypass the requirement but get flagged for administrative review.

Read that against Glenrock's log. Glenrock's own policy has required a case number on every query since 2023. Five of 375 have one. Flock is now going to enforce, by software, the rule Glenrock wrote for itself and did not follow.

3. Audit Assistance becomes mandatory for all law enforcement customers by the end of 2026. The feature analyzes user activity and flags abnormal searches for an administrator to review. It launched around April 2026 as an option; Flock says over a third of customers have adopted it.

Here's the local hook: look at the produced audit-log screenshots. Four pages of Glenrock's production are captures of Flock's Organization Audit interface, and the tab bar on each reads "Dashboard | Audit History | Audit Assistance (1)." That parenthetical one appears on every screenshot the town produced. What was flagged? Who reviewed it? What came of it? The town's own document raises the question.

4. Offense filtering for sharing. Cities will be able to choose which categories of offense other agencies may search their cameras for — Flock's own published example is allowing another city to search for a stolen vehicle, a missing person, or a violent crime while blocking searches related to immigration enforcement. That is a concrete, available answer to "could ICE query Glenrock's cameras," and the Council can direct that it be switched on.

5. Proactive lockouts that suspend an account showing abnormal behavior before an administrator reviews it; mandatory multi-factor authentication; an independent security review; and a request that every agency activate its public Transparency Portal.

Not everyone is impressed. The Institute for Justice called the package window dressing, noting that one word appears nowhere in it: warrant. That criticism is fair, and it's the right frame for Tuesday. Flock's changes are settings. They are not law, they are not local policy, and most of them are optional for Glenrock unless the Council chooses otherwise.

Which is precisely what makes them useful. Twelve days before this meeting, the vendor handed the Town Council a menu of things it can adopt Tuesday night, in public, on the record: seven-day retention, case codes now rather than in December, Audit Assistance switched on, offense filtering configured to block categories the town doesn't want. Every one of those is a yes-or-no question a councilmember can ask and a chief can answer. There is no longer a "we'd have to see if that's possible" available on any of them.

THERE IS AN EXIT — AND THE CLOCK IS SHORTER THAN ANYONE THOUGHT

This is the part residents should know before Tuesday, because it changes the conversation from complaint to decision. It is also the section most changed by the Order Form.

The contract renews automatically unless either party gives thirty days' notice before the end of the current term (§ 7.1, GLENROCK_00010). The renewal length is "the greater of one year or the length set forth on the Order Form."

The Order Form sets the renewal term at twenty-four months (GLENROCK_00020).

That is not a small distinction. It means the decision facing this Council is not "renew for a year and revisit." Miss the notice window, and Glenrock is committed through roughly 2028.

When does the window close? The Order Form sets a twenty-four-month initial term, and § 7.1 provides that the term commences upon first installation of the hardware unless the Order Form says otherwise. The Order Form does not say otherwise. The first logged search in the audit trail is October 15, 2024. Invoice INV-76999, dated October 17, 2025, reads "Year 2 of 24 Month Term, 2025 - 2026" (GLENROCK_00031).

Every one of those points to an initial term ending in the fall of 2026 — which would put the thirty-day non-renewal deadline somewhere around September 2026, weeks after this meeting.

I am not going to state that date as a fact, because the contract keys it to the installation date and the town has not published one. That is the single most actionable question anyone can ask on August 25: On what date did the term commence, and what is the last day to give notice of non-renewal? The Chief can answer it from his own files in thirty seconds. Nothing else on the agenda matters as much, because everything else can be revisited and this cannot.

Separately, § 11.15 lets Glenrock terminate for non-appropriation — simply declining to budget the money — on thirty days' written notice, "without penalty or other cost" (GLENROCK_00017). If the town walks away, Flock removes its hardware and restores the site to its original condition (§§ 7.2, 10.5, GLENROCK_00010, 00014).

Glenrock would not be unusual in walking away. Advocacy trackers count more than eighty jurisdictions that have canceled ALPR contracts or taken the cameras down; South Portland, Maine voted to cancel effective immediately in June; the Los Angeles Police Department declined to renew a contract covering 138 cameras in July.

The Council does not need to prove misconduct to decline to renew a contract. It only needs to decide the trade isn't worth it — and it should know that the trade, this time, is for two years.

QUESTIONS TO BRING TO TOWN HALL ON AUGUST 25

Print these. Hand them out. Ask them on the record.

On the contract and the clock — ask these first

1. The Order Form sets a 24-month renewal term. On what date did the initial term commence, and what is the last day to give notice of non-renewal? If the answer is September, say so tonight.

2. Did the Town Council ever formally vote to authorize this contract, or was it signed by the Chief alone? Ordinance No. 749 is a budget ordinance and doesn't mention it.

3. The Order Form was produced as an unsearchable image (GLENROCK_00019–00022). Will the town post a text-searchable copy?

4. The contract total is $34,450 over the initial term. Was that figure ever presented to the Council in public?

5. The Order Form's fee lines show three standard installations and two MASH-tested poles. Which three cameras are on existing structures, and where are the two new poles?

6. Glenrock's subscription includes "Community Network Access" — the ability to request feeds from privately owned Flock cameras at businesses, schools, and neighborhoods. Has Glenrock ever used it? How many private cameras are reachable?

7. The subscription includes "Unlimited Custom Hot Lists." Who can create one, and what lists exist today?

8. The contract grants Flock a perpetual license to use anonymized Glenrock data to train machine learning algorithms (§ 4.3). Did the Council know that? Can the town audit the anonymization, or revoke the license?

9. § 1.10 defines the services to include audio detection. The town has not bought an audio product. Will the Council commit, in writing, that adding one requires a public vote?

On the audit log

10. The policy requires every query to include "the authorized law enforcement justification for query or access, including a relevant case number." Out of 375 searches in twenty months, five carry a case number, and all five are the same case. Why?

11. Who actually reviews these audit logs, how often, and has any review ever been documented in writing? Can the Council see one?

12. Ninety-eight searches — 26.1% — are logged only as "Test" or "Training." Ninety-two are the Chief's. Section E.1.f authorizes training for new users; Section H.1 requires training before access is granted. Who was being trained in December 2025, May 2026, and June 2026?

13. On December 8, 2025, a "Hit and Run" search for a white Ram pickup at 9:39 a.m. was followed at 11:48 a.m. by forty-eight searches logged "Test," six filtering for a Ram, using windows that included the same incident morning. Were those part of the hit-and-run investigation? If so, why weren't they logged under it?

14. Entries in 2026 read "Gambling - Test" and "Trespass - Test." Was the town investigating gambling and trespass, or was the offense category false? Which field is wrong?

15. The audit screenshots the town produced show "Audit Assistance (1)." What was flagged, when, who reviewed it, and what was the outcome?

On scope, reach, and outside access

16. The town has five cameras. Where exactly are the five poles? The records request asked; the town pointed to the vendor's website. If the locations are sensitive, on what legal basis — given that the Chief has described them publicly and WYDOT permit records are public?

17. Why does a town of 2,400 need searches reaching 5,794 and 6,281 camera networks? Who authorized them, and does the Chief review multi-network searches before or after they run?

18. The Order Form says nationwide network access is "License Plate Lookup Only." Does that mean a specific plate was entered on every multi-thousand-network search? If so, whose, and under what justification?

19. On December 6, 2025, a search logged only as "Test" reached 46 networks. What was that?

20. Has any outside agency — including ICE, CBP, the FBI, or any out-of-state department — ever requested or accessed data from Glenrock's cameras? Would the town know if they had?

21. Flock now offers offense filtering, letting a city block other agencies from searching its cameras for specified purposes, including immigration enforcement. Will Glenrock turn it on, and with which categories blocked?

22. The contract lets Flock disclose footage to third parties on a "good faith belief" it's necessary (§ 5.3). Has Flock ever done so with Glenrock's data, and would the town be notified?

On retention and data control

23. On November 17, 2025, an officer searched a 108-day window against a database the Order Form says holds thirty days. Does any Glenrock or Flock-held data survive past thirty days, anywhere?

24. Flock cut its recommended default retention to 7 days on August 13 and gives Evidence Mode free to agencies that adopt it. Glenrock's Order Form specifies 30. Will the town opt in to seven days? If not, why is it paying extra to keep data longer?

25. When an officer downloads an image into a case file, the thirty-day purge no longer applies. What is the retention policy for downloaded plate images, and where are they stored?

On authority and oversight

26. The Order Form was executed July 24, 2024, and the term commences on installation. The first logged search is October 15, 2024. Were the cameras live before October, and if so, where are those logs?

27. The LPR policy is dated March 1, 2023 — sixteen months before the Flock contract. What system did it originally govern, and why was it revised on May 26, 2026? Will the town release the pre-revision version?

28. The policy was posted publicly on the town website only in August 2026, after this records request. What notice did residents get in 2024?

29. Glenrock's own state legislator says he learned the cameras existed about a year after they went up. What does the town accept as evidence that residents knew?

30. An alert was generated for "Sex Offender 133195." The policy's Hot List definition covers wanted and missing persons, not registrants. Who added that category, and what other hot lists is Glenrock subscribed to that aren't in the policy?

31. Flock will require case codes on every search by the end of 2026. Will Glenrock require them starting Tuesday?

32. The vendor's transparency portal appears nowhere in the contract and nowhere in the policy. The town has no enforceable right to it. Will the Council write a public-reporting requirement into the renewal or into policy by ordinance?

33. What independent oversight exists, other than the Chief auditing himself? Would the Council support a civilian review requirement, an annual public audit, or a warrant requirement for searches beyond a set window?

On Rolling Hills and on the decision itself

34. Rolling Hills residents are policed under this system but don't vote in Glenrock or sit on this Council. Were they consulted, and do they have any say in a two-year renewal?

35. Section 11.15 allows termination for non-appropriation with no penalty. What would the town lose, concretely, by declining to renew for one year while it writes stronger rules?

36. Wyoming DCI's own data shows Glenrock's reported property crime rose from 7 offenses in 2024 to 12 in 2025 — the first full year with the cameras — and larceny went from 3 to 10. Does the department dispute those figures? If not, what is the case that this system is working?

37. Of the property crimes reported in 2025, how many were cleared using Flock? Not searches run — the log already shows 375. Cases closed. And what was the clearance rate in 2022 and 2023, before the cameras? It was 58.33% in 2025.

38. What would have to happen for this Council to cancel this contract? If nothing would, what is the point of an informational meeting?

A NOTE ON METHOD, AND ON WHAT'S INFERENCE

Every count in this piece was computed, not estimated. I extracted all 375 rows of the audit log into a spreadsheet, converted every timestamp from UTC to Mountain Time, and ran the totals. Some pages of the log carry no text layer at all; I read those visually. Where the article gives a figure, it came out of that spreadsheet, and you can rebuild it from the Bates-stamped pages yourself.

The lesson of this correction is worth stating plainly, because it applies to anyone who reads a records production. Pages GLENROCK_00019–00022 have no text layer. Search the PDF and they come back empty. I concluded from that emptiness that the Order Form had not been produced, and I published that conclusion. The pages were legible the entire time; they simply had to be looked at rather than searched. If you are checking my work — and you should — open the images, don't just run a keyword search. A document that a computer cannot read is not the same as a document that does not exist, and I should have known the difference before I wrote the first version.

Some things in here are inference, and I want to mark them clearly.

That the December 8 "Test" searches were a continuation of the hit-and-run investigation is my reading of the pattern — matching vehicle make, overlapping time windows, hours apart. It is not established fact. One detail cuts against my own reading: the first four Ram-filtered searches used the wide four-day window and only afterward narrowed to the incident morning, which is not the order you'd expect from a straightforward continuation. There is an innocent explanation — an officer running unrelated system practice who happened to type a common truck make. I think the pattern is more than coincidence. Chief Felton is the person who can say for certain.

That the initial term ends in fall 2026 is inference from three consistent data points — the 24-month term, the October 2024 first search, and the invoice's "Year 2 of 24 Month Term, 2025 - 2026." The contract keys the term to the installation date, which the town has not published. Ask for it.

That two cameras sit on new MASH-rated poles and three on existing structures is inference from the fee lines on the Order Form. It is a reasonable reading, not a disclosure.

The license plate column was redacted, so for some searches I cannot rule out that a specific plate was entered. Where I describe searches as having no vehicle description, I am describing the record as produced.

My description of how most residents learned about these cameras is an observation, not a survey. That is why I've suggested testing it in the room rather than asserting it as a finding.

The crime figures come from Wyoming DCI's published agency data. The 2025 case counts are read directly; the 2024 property figures are derived from the percentage changes the portal displays. The three derived numbers are mutually consistent, but they are calculated, not printed.

And most of what's in this log is legitimate police work. A missing juvenile in November 2025. Felony warrants. Stolen vehicles. DUIs. An animal cruelty case. If the cameras helped find a runaway teenager, that matters, and no argument here says otherwise.

The argument is narrower, and harder to dismiss: a tool this powerful is only as safe as the record that shows how it was used. Glenrock's record does not show that. It is worth fixing before the question stops being hypothetical — and it is worth fixing before a renewal that runs two years.

I'll be reporting on what's said on the 25th, and on whatever the town's attorney produces next.

WHAT CHANGED IN THIS VERSION

The Order Form is not missing. Previously reported as absent from the production; it is at GLENROCK_00020, with the billing schedule at 00021 and the product description at 00022. The pages are image-only and return nothing to a text search. Every reference to a missing or blank Exhibit A has been removed.

Renewal term corrected from one year to twenty-four months (GLENROCK_00020), and the "exit" section rewritten around it.

Contract total corrected from "roughly $30,000" to $34,450, with $19,450 due at signing.

Retention confirmed at thirty days on the Order Form, resolving the open question under MSA § 1.14.

New section on what FlockOS includes — nationwide and state network lookup, law enforcement network access, community network access to privately owned cameras, real-time NCIC alerts, unlimited custom hot lists (GLENROCK_00022).

The audio point sharpened: the Order Form shows the town bought LPR only, no audio product. The contractual authority remains broader.

The hot list section now notes that unlimited custom hot lists is a purchased feature.

Questions renumbered and expanded from 30 to 38, reordered to put the renewal deadline first.

SOURCE DOCUMENTS

The complete production is posted here: The complete Glenrock Flock production. Read it yourself; every Bates number below is stamped on the page.

All Bates-numbered citations are to the Town of Glenrock's August 19, 2026 production under the Wyoming Public Records Act, in response to a July 17, 2026 request:

Master Services Agreement with Flock Group, Inc. — GLENROCK_00001–00025

Definitions incl. § 1.10 Flock Services, § 1.14 Retention Period — 00002–00003

§ 4.1 Customer Data, § 4.3 Anonymized Data — 00006–00007

§ 5.3 Disclosure of Footage — 00008–00009

§ 7.1 Term and renewal, § 7.2 Termination — 00010

§ 9.1 Limitation of Liability — 00012

§§ 10.1–10.5 Installation and hardware ownership — 00013–00014

§ 11.8 Publicity, § 11.15 Non-Appropriation — 00016–00017

Exhibit A / Order Form — 00019–00023 (cover 00019; Order Form 00020; billing schedule 00021; product description 00022; execution page 00023)

Exhibit B / Insurance — 00024–00025

Ordinance No. 749, FY 2024–25 budget — GLENROCK_00026–00028

Expenditure Budget Report 2024-2025 — GLENROCK_00029

Flock payment information — GLENROCK_00030

Invoice INV-76999 (10/17/2025) — GLENROCK_00031–00032

Flock W-9 (2025) — GLENROCK_00033

Glenrock PD Policy No. 10.02, License Plate Readers, issued 03/01/2023, revised 05/26/2026 — GLENROCK_00034–00039

Audit Logs, October 2024 – June 2026 — GLENROCK_00040–00074

Policy also posted at glenrock.org (uploaded August 2026). Vendor transparency portal: transparency.flocksafety.com/glenrock-wy-pd. Meeting details from the Town of Glenrock calendar; officials via glenrock.org/mayor-and-council.

REPORTING AND LEGAL SOURCES

Cowboy State Daily, "Unlike Other Areas In Wyoming, Glenrock's Flock Cameras Bring Little Drama" (Clair McFarland, July 3, 2026) · Cowboy State Daily, "License Plate Reader Map Right That Lusk Has Them, Wrong About Cody And Gillette" (July 15, 2026) · Chatrie v. United States, No. 25-112, 609 U.S. ___ (June 29, 2026) · Carpenter v. United States, 585 U.S. 296 (2018) · Wyo. Const. art. 1, § 4 · Wyo. Stat. Ann. §§ 16-4-201 to 16-4-205 · Truthout, "SCOTUS Ruling Could Be Bad News for Flock — But Won't Stop Mass Surveillance" (Mike Ludwig, July 10, 2026) · EFF, "How Cops Are Using Flock Safety's ALPR Network to Surveil Protesters and Activists" (Nov. 3, 2025) · EFF, "License Plate Surveillance Logs Reveal Racist Policing Against Romani People" (Nov. 2025) · EFF, "Flock Safety and Texas Sheriff Claimed License Plate Search Was for a Missing Person. It Was an Abortion Investigation." (Oct. 2025) · ACLU, "Flock Safety: Credibility Lost" (July 2, 2026) · Flock Safety, "Flock Updates Privacy, Accountability, Security, and Transparency Safeguards" (Aug. 13, 2026) · Institute for Justice, "Institute for Justice Calls Flock Safety's Proposed Reforms 'Window Dressing'" (Aug. 2026) · Forbes, "Flock Installed AI Cameras In This Small City And Claimed Crime Went Down. It Went Up." (Cyrus Farivar, Feb. 29, 2024) · Wyoming Division of Criminal Investigation, Crime Insight portal, crimestats.wyo.gov · Wyoming DCI, Uniform Crime Reporting/NIBRS, wyomingdci.wyo.gov · DeFlock.org