Reporters Found What Police Audits Missed

On August 28, 2026, USA TODAY published the largest independent look yet at how police actually use Flock’s license-plate-reader network. Reporters analyzed more than 75 million searches, associated with about 6,300 agencies across 24 states, over three and a half years. What they found is not a story about catching criminals. It is a story about who was watching the watchers — and the answer was often not the police departments themselves.

In case after case, USA TODAY reporters spotted the irregular activity first. Only after a reporter called did the agency act. At least 19 law-enforcement employees in five states were fired, suspended, charged, or placed under investigation in a two-month span. In Georgia, the state bureau of investigation has announced 14 arrests since July 6 — several identified with the help of Flock’s own auditing tools. A 911 director searched her estranged husband and his new girlfriend more than 2,500 times. A Texas officer searched seven plates — one more than 3,400 times — and now faces at least 100 felony counts. A Georgia detective his department believed had made 41 searches of an ex-partner had, in the full record, made 2,890.

The tool that made this possible is the same one we have spent months fighting to obtain in North Carolina: the audit log. Until December 2025, those logs showed each searcher, the plate searched, and the stated reason. Then, on December 9, 2025, a Flock vice president emailed customers that the company would begin blacking out most of it — officer names, the plates searched, vehicle “fingerprints,” and the open-text reasons — so that an agency could see only its own users; searches run by outside agencies on its cameras would read “REDACTED” or simply “***”. The Electronic Frontier Foundation’s Dave Maass called the change “catastrophic” — not just for the public, but for agencies’ own ability to catch outside abuse. Flock said it acted to protect undercover officers; asked for a single example, the company provided none.

And even USA TODAY’s 75-million-search database is badly incomplete: roughly three-quarters of officer names and license plates are unavailable in the source records. What reporters found was only what remained visible after the redactions — not necessarily everything that occurred.

Read this first — where the North Carolina data came from.
None of the North Carolina search activity in this investigation was released by North Carolina. North Carolina makes captured license-plate data confidential and not a public record (G.S. § 20-183.32(e)), and our agencies decline to disclose their own Flock activity. Yet traces of how North Carolina officers search the nationwide Flock network still appear in the audit records produced by agencies in other states — because a North Carolina search that reaches out-of-state cameras is logged there too, in jurisdictions with different public-records laws. In other words, a North Carolina agency’s refusal to disclose its own activity does not keep that activity from becoming visible elsewhere in the network. The secrecy may limit public oversight here more effectively than it limits the information moving through the network itself.

Why the redaction doesn’t fully work — and how we read the Cape Fear

There is a reason the December blackout came too late. Because a Flock search can reach cameras belonging to many other agencies through shared networks, traces of the same search can appear in the audit logs of multiple jurisdictions it touches — even when the searcher’s own department withholds its records. That is how USA TODAY reconstructed a picture of 208,000 officers, and it is the same principle behind our own recent work matching two North Carolina productions back together.

Using the public audit data — the same records USA TODAY drew on, and, to repeat the point, records North Carolina’s own agencies would not release; they are visible only because other states did — here is what the Cape Fear region looks like. One caution first, and it matters: a high search count is not proof of wrongdoing. A hot case, a wanted vehicle, or a multi-agency investigation can all produce heavy, legitimate searching. USA TODAY says this plainly, and so do we. What the numbers show is where questions are worth asking — not answers.

Brunswick County. The Sheriff’s Office accounts for roughly 11,014 searches in the records. One stands out: a single license plate searched 418 times on five separate days (spanning June to September) by one user — the entire search history of that plate coming from that one account. We are not identifying anyone or alleging misconduct, and we do not know whether this pattern is related to, or separate from, the office’s recently disclosed personnel matter now under State Bureau of Investigation review — we simply cannot tell from the public data. But a single-user, single-plate concentration is precisely the pattern a misuse audit is built to flag — and Brunswick’s own Sheriff has said an internal audit is what caught the matter he referred to the SBI.

Pender County. About 2,336 searches, with most of the high-frequency activity tied to one investigator. Here the fuller picture cuts the other way: a second Pender officer — and a State Bureau of Investigation agent — searched the same plates in the same window. Several agencies working the same tags reads like a joint investigation, not personal abuse. It is a useful reminder that the data rewards careful reading, and that a high count, by itself, proves nothing.

New Hanover County. New Hanover ran 15,963 searches in this record — the most of any Cape Fear agency — across 593 different plates, with three users accounting for much of it (one alone ran 951 searches across 180 plates). Its most common stated reason was “Drugs/Narcotics – Vice.” Yet no single plate is hammered hundreds of times the way the state’s outliers are: New Hanover’s most-searched plate tops out around 60. That is a different profile — broad, heavy use rather than a single-plate fixation. It raises a different question, too: not “is someone stalking one driver?” but “how extensively is this surveillance capability being used at all?” — a judgment a public audit lets a community make.

Carolina Beach. The Town’s roughly 4,988 searches are spread thin, with no single runaway pattern — genuinely restrained by comparison. Heavy searching is not universal, and it is not inevitable.

The Cape Fear region, side by side. Each figure below is the single most-concentrated plate in that agency’s records — the one plate searched most often, and how much of that plate’s activity came from a single user. Totals are searches run by that agency’s officers, across the full window the public data covers (for New Hanover, Nov. 2024–June 2026); the per-plate figures fall within it. Throughout, “days active” means the number of separate days a user searched that plate — not a continuous stretch — so a plate searched on 5 active days may span several months.

For contrast, the statewide extremes are in another league.

Frequency band and score are USA TODAY’s metric, combining how often a plate was searched, over how many days, and how concentrated it was in one user. A high count is not proof of misuse — it marks where questions are worth asking. The single Gastonia user above appears atop the statewide list more than a dozen times; Robeson County recurs through the rankings as a cluster of separate heavy users.

In Gastonia, one user searched a single plate 593 times over four months — the highest frequency score in North Carolina — and appears atop the state list more than a dozen times over. In Robeson County, one user searched a plate 582 times across three active days. These are the kinds of concentrated patterns that deserve examination — the same kind that, elsewhere, uncovered searches investigators ultimately deemed improper.

Two sheriffs, two answers

This is not only a national story or a data story. It is a question of how a sheriff chooses to treat the public — and on our own coast, two of them have answered it differently.

Brunswick County shows how disclosure can work. When a routine Flock audit flagged a problem, Sheriff Brian Chism made it public, the employee resigned, and he referred the matter to the State Bureau of Investigation. North Carolina law required none of that. He did it anyway.

New Hanover County’s answer has been more guarded. On August 25, 2026, responding to Commissioner Stephanie Walker, Sheriff McMahon said the office was not aware of any misuse and announced safeguards: monthly audits he would personally review, cutting data retention from 90 days to 30, two-factor authentication, individual user accountability, and continued work on a public transparency portal. Those are meaningful steps, and worth crediting. But they leave the central accountability question unanswered. McMahon has indicated the public would learn of misuse chiefly if it results in criminal charges — and under North Carolina’s personnel-records law, that is largely how it works: most discipline short of dismissal stays sealed. So if a monthly audit turns up improper use that ends in internal discipline but not prosecution, what, if anything, will the public ever be told? That is a structural transparency gap, not a personal failing — and USA TODAY’s reporting shows why it matters: in case after case, departments had not caught the conduct until a reporter called.

That is why independent oversight — by reporters, by researchers, by residents reading public records — is not a nuisance. Where the system is built so the public generally learns of abuse only when it becomes a criminal case, outside scrutiny is much of the scrutiny there is. And it is why the design of the system matters more than the character of any one deputy: a surveillance network should be judged not by how the best officer uses it, but by how the worst one could. A nationwide, lightly-audited plate-tracking network hands that worst actor enormous reach. We’ve made this case at length in Design for the Worst Sheriff.

The point for our community

Two things follow. First, the oversight that surfaces this misconduct has often come from reporters, researchers, and residents reading public records — not from the agencies running the cameras. Some departments and Flock’s own auditing tools have caught misuse too, but frequently only after the public pressure this kind of reporting created — and Flock’s response was to make the records harder to see, not the abuse rarer. Second, North Carolina’s confidentiality law creates an unusual asymmetry: residents here generally cannot inspect the records needed to audit how ALPR surveillance is being used, while traces of North Carolina officers’ searches may still surface when they cross into states with different public-records laws. The secrecy may limit public oversight more effectively than it limits the information moving through the network.

That is exactly why the decisions in front of our own towns matter. On September 8, 2026 the New Hanover County Commissioners voted 4-0 to cancel the county’s Flock contract — but the beach towns and neighboring counties still run their own. Where the public can speak next, and the petition, are how this keeps moving.

• Read the Carolina Beach story: deflockilm.org/flock-cameras-carolina-beach
• The running record of NC Flock misuse: deflockilm.org/nc-flock-misuse
• Sign and share the petition: change.org/DeFlockILM
• Email your elected officials: deflockilm.org/email-your-leaders

Explore the data: who searched, and what

The audit data also names the most active users at each agency — by initials, as the public tool shows them. These are not accusations. A detective assigned to narcotics or auto theft will run large volumes of legitimate searches; heavy use is what several of these officers are supposed to do. What the records let the public see is simply who is using the network, and how much.

A selection of the highest-frequency plates run by those users appears below (plates redacted at the source; a sample of each user’s top records, not their full activity). Two rows are worth watching: the single Brunswick plate run 418 times, and the Carolina Beach cluster of eight different plates each searched once on the same day — a batch lookup rather than a sustained watch.

Not an accusation of wrongdoing. Frequent searching can have legitimate law-enforcement reasons; USA TODAY says so, and so do we. The value of the public audit is that a community — and a supervisor — can at least see the activity and ask. Except where a full per-user total is shown, figures describe a user’s top records, not their entire activity.

Sources: USA TODAY, “Our Flock camera investigation sparks police firings, arrests” (Aug. 28, 2026, updated Sept. 3, 2026) and its public Flock search database; Flock Safety’s December 9, 2025 customer notice; WECT (Sheriff McMahon’s August 25, 2026 responses to the New Hanover County Board of Commissioners); N.C. Gen. Stat. § 20-183.32. Search figures are drawn from a selective review of publicly released audit data and are current to late 2025/early 2026; a high count is not, by itself, evidence of misuse. General information for public discussion, not legal advice. Corrections: mark@deflockilm.org.

Related

• Who is searching North Carolina’s Flock cameras — and from where.
• The running record of North Carolina Flock misuse.
• Flock cameras in Carolina Beach.