Technology Should Never Operate Without Accountability. The Sheriff Said It. Now Apply It.

“Technology should never operate without accountability.” — Sheriff Ed McMahon, August 10, 2026. We agree. So let’s apply the test he just wrote — starting with his own portal.

Credit where it’s due — and to the Sheriff specifically

A week before the County Commission takes this up, Sheriff McMahon announced three real changes to the Sheriff’s Office’s Flock program: retention cut from 90 days to 30; a monthly internal audit, up from an annual one; and a public Flock transparency portal with camera counts, search totals, the policy, and recent audits.

These are not nothing, and we don’t treat them as nothing. They make the system less dangerous, and they exist because residents signed, spoke, filed records requests, and refused to let it pass in silence. If you were one of them, this is partly your doing.

And genuine thanks to Sheriff McMahon. We mean that plainly, not as a setup. He engaged. He cut retention, opened a portal, and stood in front of a camera to answer for the program when he did not have to. That took some measure of good faith, and we credit it.

Which is exactly why Wrightsville Beach should be ashamed. Set the two side by side. The Sheriff moved; the Town of Wrightsville Beach stonewalled. Thirty-six-plus days, not one Flock record produced, the request shipped off to outside counsel, and public silence — an F on our transparency scorecard, and earned. A town whose own police department lists “nurturing the public trust” among its values has produced nothing and answered no one. If the Sheriff — running a far larger program — can cut retention and post a portal in the same window, Wrightsville Beach’s stonewall is a choice, not a constraint.

The Sheriff has done more than every local agency around him. That is worth saying plainly. It is also why the gaps that remain deserve a straight answer — and why the portal, the centerpiece of his response, has to be held to what it claims to be.

The Sheriff’s statement, in his own words

“The New Hanover County Sheriff’s Office is committed to protecting our community while maintaining the public’s trust.”

“License plate reader technology has become an important investigative tool. This technology helps us solve crimes, find people in danger and keep our community safe. But it must always be used responsibly.”

“We understand with technology comes an expectation of transparency and accountability. And that’s why we’re taking additional steps to strengthen public confidence in how the system is being used.”

On the balance he says he is trying to strike: “…the right balance between protecting public safety and protecting the public’s trust” — and “we believe these additional measures strengthen both.” On misuse: “Abuse of this technology will not be tolerated,” with employees held “fully accountable,” up to termination and criminal prosecution. And the line that frames this whole piece: “Technology should never operate without accountability.”

How it was delivered matters too. He read the roughly three-minute statement, answered one question — whether the portal was live yet — and when asked why he wouldn’t take further questions, removed his microphone and returned to his office. WECT had sought an on-camera interview about Flock for two months.

His standard, applied

The Sheriff didn’t just announce policies. He named a standard: trust, accountability, and “the right balance between protecting public safety and protecting the public’s trust.” We take him at his word — and hold the rollout to it. The profession wrote that standard down:

“Trust between law enforcement agencies and the people they protect and serve is essential to the stability of our communities, the integrity of our criminal justice system, and the safe and effective delivery of policing services.”
— President’s Task Force on 21st Century Policing (2015), the opening line of Pillar One, “Building Trust and Legitimacy.”

The Task Force’s remedy is about sequence: engage the public before a surveillance technology goes up, not explain it after. Measured that way, August 10 has the sequence backwards. A standing, warrantless, county-wide camera network went up first; the public deliberation never happened; and the “additional steps,” however welcome, are again the office’s own unilateral decision, announced to a TV camera. Accountability the accountable party defines, grants, and can revoke by itself is a weaker thing than it sounds — and it still leaves the public without the three facts that decide the question: what the cameras are, where they are, and who has searched them. (More on the trust standard.)

What the cameras actually are

Has his position changed? On what the cameras are, no. In his July memo to the Commissioners, the Sheriff described “License Plate Reader technology” — cameras that “photograph the rear of vehicles only” and collect “limited vehicle-related information.” On August 10 he said the same: an investigative “license plate reader” tool, not a live surveillance system, collecting only “limited” vehicle information.

His own portal tells a different story. The top line of its Data & Usage Summary reads: Total cameras: 36. License plate readers: 19. That leaves 17 cameras that are not license plate readers. The county’s signed contract lists eight Solar Condor PTZ units — dual-lens, pan-tilt-zoom cameras that record video and can zoom in on people. We have already located and mapped five of the eight Condors — and one of the integrated Axis cameras: one watches a new neighborhood, another points toward a children’s soccer field at Veterans Park. So the Sheriff’s own portal counts more than double the video cameras we already knew about — 17, not 8 — on a system he still describes to the public as reading “the rear of vehicles only.” What the remaining cameras are, the portal does not say — and neither has the Sheriff.

A “transparency portal” that isn’t transparent

Standing up a public portal only after being called out is one thing. Standing up a portal that calls itself transparency while omitting the facts that matter is worse. Six things the portal does not disclose, all evident from its own pages:

• It counts more cameras than it explains. The portal lists 36 total cameras and 19 license plate readers. That leaves 17 the portal never identifies. We can account for some: the county’s invoice shows eight Solar Condor PTZ video cameras in the contract, and we have located five. What the remaining cameras are, the portal does not say — and neither has the Sheriff. That gap is a factual question the county has left open.
• The search count quietly measures only half the system. “Number of Searches: 815” counts only the Sheriff’s Office’s own searches — the organization audit. It does not include the Network Audit: the far larger log of outside agencies querying the shared national network New Hanover’s cameras feed into. No disclaimer, no footnote.
• The downloadable CSV carries the same blind spot. The file the portal lets you download is that same organization audit — the county’s own searches only, the half of the story it already controls.
• 774,174 vehicles logged in 30 days. The portal reports that many “Unique Vehicles Detected.” Each is captured, time-stamped, and made searchable — the overwhelming majority belonging to residents who are not suspected of anything.
• “What’s detected” answers a question no one asked. The portal says the system detects “License Plates, Vehicles” and not “Facial Recognition, Gender, Race.” Keep the distinction precise: a lens that records a face is not the same as software that detects, indexes, and stores faces as searchable data — and we are not claiming Flock does the latter. But the disclaimer dodges the real question. These cameras still record people: the Condor PTZ units capture video of individuals, not just plates (watch what these cameras actually do). And New Hanover’s own records show the system used to search for a person by description — including a racial term: the search was blocked by Flock’s filter, then re-run fifteen seconds later with the race word removed and allowed. And because these cameras produce ordinary video, that footage can be exported and run through a separate facial-recognition service — as 404 Media has documented — so the ability to put a name to a face belongs to whoever holds the video, whether or not Flock offers it. “We don’t run facial recognition” is not the same as “these cameras can’t identify you.”
• “Valid law enforcement purpose” is contradicted by the county’s own contract. The Access Policy says every search “requires a valid law enforcement purpose” and is “stored indefinitely.” Yet the Network Audit volume shows Texas, Georgia, Florida, and federal agencies reaching into these cameras — none bound by North Carolina’s law-enforcement-only limit — and the county’s own signed contract permits Flock to disclose footage to “law-enforcement authorities, government officials, and/or third parties” on a mere “good-faith belief” (p. 14). The portal states a rule the county’s own paperwork does not keep.

Note the tell in the fine print: the captures are deleted in 30 days, but the search records are stored indefinitely — and those are exactly the outside-agency records the county still refuses to show. The portal keeps the log it won’t disclose and deletes the data that might have exonerated you.

And the public cannot even archive it. When we tried to preserve the portal on the two public web archives — the Internet Archive’s Wayback Machine and archive.today — both were turned away with an HTTP 403. The page sits behind a Cloudflare bot-wall — possibly a default security setting rather than a deliberate choice — that nonetheless serves the portal only to a live human browser and turns away the archives built to preserve public records. Whatever the intent, the effect is the same: a transparency portal the public cannot keep its own copy of. We saved ours by hand.

The monthly audit watches the wrong searches

A monthly internal audit reviews the Sheriff’s Office’s own searches — 815 in the last 30 days. But the deputies are not who mostly searches this data. Here is the mechanism that matters: New Hanover chose to share its cameras into Flock’s nationwide network, so when an agency anywhere runs a search, that query sweeps every network it can reach — New Hanover’s included — and Flock’s audit records it as a search of New Hanover’s cameras. Outside agencies aren’t tapping our live sensors one by one; they are querying the shared national system our plate captures feed, and getting our data back. In the five unredacted North Carolina productions we hold — sixteen monthly Network Audit files, deduplicated on Flock’s own unique search IDs to 3.88 million distinct searches by 3,904 agencies over three months — about 90% came from outside North Carolina (state assigned from each searching agency’s name). The single largest searcher was Houston, Texas Police; Texas, Georgia, and Florida lead the list.

New Hanover redacted the one field that would show who searches its cameras, and that record is still withheld; we filed for pre-litigation mediation over it on August 3. So the monthly audit, however diligent, cannot see roughly nine in ten of the searches touching New Hanover’s cameras — run by out-of-state agencies the Sheriff has no authority to audit, whose copy of your plate data lives under their state’s looser rules once it crosses over. Cutting our retention to 30 days does nothing to the copy that already left the state. The portal even concedes the mechanism: it says data is shared only when “your agency decides who to share your data with.” Precisely — New Hanover’s agency decided. (Who is searching North Carolina’s cameras.) (The records timeline.)

“Abuse will not be tolerated” — caught how?

Take the pledge as sincere, then ask the mechanical question: caught how? The audit that is supposed to catch misuse is the same audit whose labels the office has already asked the public to discount. When WECT reported that New Hanover’s records logged searches coded “Traffic Infraction” — a purpose N.C. Gen. Stat. § 20-183.31(b) says ALPR data “shall not be used” for — the office’s answer was that they were “mislabeled.” But you cannot hold up the audit as the safeguard and, in the same month, wave off its contents as unreliable. If “Traffic Infraction” doesn’t mean a traffic infraction, the audit doesn’t reliably record why any search was run — and the one accountability mechanism everyone points to is broken. Either the label means what it says, or the safeguard means nothing. (“Mislabeled”? Then fix the labels.)

And even a real catch stays invisible. North Carolina keeps law-enforcement personnel and disciplinary records confidential. Unless an officer is criminally charged — as in Mooresville and Charlotte — residents never learn an audit caught anything. “Fully accountable” operates behind a wall the public isn’t allowed to see behind.

The portal’s search log, decoded

Take the portal’s download at face value — that same organization audit, 815 of the county’s own searches over 30 days — and decode it. Even limited to the half of the system the county controls, it makes the case better than we could.

• The dragnet is real, in the county’s own numbers. Every routine local search fanned out across a median of 2,417 camera networks (max 6,515); 80% reached 2,000+. One deputy looking up one car sweeps thousands of outside networks.
• It’s mostly drugs, not missing children. The system is sold on missing kids and violent crime. In the data, 54.6% of searches are Drugs/Narcotics. Homicide is 0.6%, robbery 0.5%, missing/endangered 0.5%.
• It records a use the law bars. 13 searches are coded “Traffic Infraction” — the very code the office calls “mislabeled.” Ten list no reason at all.

Better rules aren’t the right question

Everything in our earlier piece, “The Camera Caught the Officer. Would Regulation Catch the Rest?”, still applies. It’s a start — but regulating a standing, warrantless, mass-surveillance system is not the same as removing it, and it isn’t enough. Tightening retention and auditing use accepts the premise that the county should be running it at all — a premise never put to the public. Thirty days of tracking innocent people is still thirty days of tracking innocent people.

What still hasn’t happened — and what August 17 is for

Strip away the announcement, and four things stand exactly where they did before August 10:

• No public vote. The system went up without one — and a statement read to a TV camera is not a public decision.
• What the cameras are is still unacknowledged. Thirty-six cameras, nineteen plate readers, seventeen others the Sheriff has never explained.
• Where they are is still secret.
• Who searches them is still withheld. The network audit sits in pre-litigation mediation, unproduced.

The people who can change that meet Monday, August 17, at 4:00 PM — Historic Courthouse, Room 301, 24 N. Third Street. Public comment comes at the end of the agenda; sign up in person by 3:30.

Hold the Sheriff to his own words. Tell them what the cameras are. Show us where they are. Name who is searching us. Then put it to a vote.

Sources

• WECT — “WECT Investigates: Sheriff addresses Flock concerns, announces policy changes” (Connor Smith, Aug. 10, 2026).
• New Hanover County SO Flock Transparency Portal (transparency.flocksafety.com), captured Aug. 10, 2026.
• DeFlockILM aggregate of five North Carolina agencies’ unredacted Flock Network Audits (3.88M distinct searches). Related: who is searching · the Sheriff’s position · the “mislabeled” searches.

This page is general information about North Carolina and federal law, not legal advice, and creates no attorney-client relationship. For advice about a specific situation, consult a licensed North Carolina attorney.