Rebuttal · Flock Safety announcement, August 13, 2026

Settings Are Not
Safeguards.

Flock announced shorter default data retention, mandatory case codes, and new audit tools. Every one of those is a setting that a company or a police chief can change back on a Tuesday with no public process. None of them is a law. None of them required your consent. Here is Flock’s announcement read line by line against Apex Police Department’s own search log.

DeFlock Apex · Published August 14, 2026 · Primary sources: Flock Safety, Apex PD search audit log, N.C. General Statutes

On August 13, Flock Safety published a list of changes to its license plate reader system. Shorter default retention. Mandatory case codes. Wider rollout of an audit tool. Offense filtering for sharing. The coverage treated it as reform.

Read it again and count how many of those changes a resident of Apex can enforce. The answer is zero. A safeguard is something residents voted for and can hold their government to. A setting is something a vendor or a police chief adjusts internally and can adjust back. Every item on Flock’s list is a setting.

That is not a rhetorical point. It is the entire problem in Apex. This camera network went up without a resident vote, without an ALPR ordinance, and without a published surveillance impact report. A company revising its own defaults does not fix a consent problem. It demonstrates it.

The question was never how long Flock keeps the data. It was who decided to collect it, and nobody asked us.

The one sentence that matters most for Apex

Buried in the retention section is a claim about how retention periods were set in the first place.

Flock Safety, August 13, 2026
“Existing customers will keep their current, democratically approved retention periods.”

Apex’s retention period is not democratically approved. It was set in a vendor contract. There has been no recorded Council vote authorizing this camera network, no ALPR ordinance, no surveillance impact report, and no public hearing on the decision to join a nationwide sharing network. Flock is describing a democratic process that did not happen here.

That same sentence has a practical consequence people are missing. The seven-day number in every headline is not Apex’s number. Apex’s Flock contract specifies 30 days. Flock says existing customers keep what they have. Unless Apex PD affirmatively opts in and says so in writing, nothing about the storage of your movements in Apex changed on August 13.

Ask the department directly: has Apex PD adopted the 7-day setting, on what date, and in what document? If the answer is no, then the reform everyone is reporting on has not reached Apex at all.

Every claim, and what it actually does

Claim 1 · Data retention
The cut to seven days is a default, not a cap.
What Flock said
“Today, we are updating our recommendation and default to a 7-day retention … Agencies can still set retention to whatever their state or local policy requires … Every community will continue to choose the retention period that fits its public safety strategy.”

What it actually does

A recommendation with an opt-out is not a limit. Any agency that wants months of location history still gets months of location history. It just has to ask. Flock is not the one unlocking the door here, but it built the door and it hands out the keys.

There may also be a price attached. The Electronic Frontier Foundation notes that Flock has previously charged customers to extend retention beyond the default, which means the real constraint on a longer window may be a line item rather than a policy. EFF calls the new default the most consequential of the changes, and also the one most easily reversed.

What it means for Apex

Apex is contracted at 30 days. Flock’s announcement explicitly grandfathers that. And the retention window was never the core problem: a single Apex search reaches thousands of outside networks and roughly a hundred thousand cameras. Storing a dragnet for a shorter time leaves it a dragnet.

Verdict: Seven days is Flock’s suggestion. Thirty days is Apex’s contract. A suggestion is not a rule, and a default is not a right.
Claim 2 · The 90 percent concession
Flock just admitted the extra three weeks were never about solving cases.
What Flock said
“Our analysis shows that over 90% of searches without a full plate are done within a week.”

Read the wording precisely

Much of the national coverage rendered this as roughly 90 percent of all searches. Flock’s own sentence is narrower than that. It says searches without a full plate. We are flagging the difference because the exact wording is the part that will get checked, and we would rather be precise than loud.

Why it still lands

Either version concedes the same thing. Flock set the default at 30 days nine years ago and defended it as a careful balance. It now says a week covers the overwhelming majority of the work. If the investigative value of this data lives in week one, then weeks two, three and four were never producing cases. They were producing inventory.

Verdict: This is the most useful sentence in the announcement, and Flock wrote it. Use their number, not ours.
Claim 3 · Evidence Mode
The trigger is a typed case number. North Carolina law already requires more than that.
What Flock said
“Evidence Mode … addresses the exceptional cases that take longer: detectives can preserve specific ALPR data as evidence for an active investigation in cold storage, consistent with existing laws and policies.”

The open question

The ACLU’s reading is that this depends entirely on scope. If Evidence Mode preserves only hit data that police have determined is evidence in a specific active case, it may be a real improvement. If it preserves any data that gets searched, it could indefinitely retain location records on people no officer suspects of anything, which would make the retention problem worse rather than better. Flock has not published the scope. That is attributable to the ACLU, not to us, and it is a question rather than an accusation.

What North Carolina law says

North Carolina does not leave preservation to a vendor’s feature design. Under G.S. 20-183.32, captured plate data may not be preserved more than 90 days, and preservation beyond that requires either a search warrant or a written, sworn statement that specifies the particular camera, the particular plate, the dates, the case and parties, and “specific and articulable facts” connecting the data to an ongoing investigation.

A case code typed into a drop-down is not a sworn statement, and it does not identify a camera, a plate, or a date range. So the question for the town attorney is direct: if Evidence Mode preserves data on the strength of a case code, how does that satisfy G.S. 20-183.32(c)?

Verdict: Flock is offering a product feature where the statute asks for a sworn filing. Those are not the same thing, and the statute wins.
Claim 4 · Offense filtering and “more local control”
It points outward, it is opt-in, and it relies on the field that already failed.
What Flock said
“Now, cities can choose which type of offenses are permissible for other agencies to access their cameras. For example, City A could allow City B to search its cameras only for a stolen vehicle, missing person, or violent crime while blocking searches related to immigration enforcement.”

Three problems

  • It faces the wrong direction. This controls what outside agencies can search Apex’s cameras for. It does not limit what Apex operators search for across everyone else’s cameras.
  • It is per agency and opt-in. Apex’s own sharing snapshots show 994 organizations in active sharing in March 2026 and 1,070 in April, 808 of them out of state. Unless every one of those agencies sets its own filters, the data still moves.
  • The filter reads a self-reported field. Whether a search is permitted depends on the offense the searching officer says he is investigating.

This has been tried

According to the ACLU’s breakdown, Flock rolled out a “Proactive Search Term Tool” on June 12, 2025 that was supposed to block impermissible searches. Officers defeated it by entering “investigation” or even “hehehe” as the reason, as reported by the Eugene Register-Guard. Flock replaced the free-text box with a drop-down menu, which hands police a list of approved answers to pick from whether or not the answer is true. Those specifics are the ACLU’s and the Register-Guard’s reporting, not our own records work.

Verdict: A permission system that trusts the requester to describe his own purpose is not a permission system. It is a form.
Claim 5 · Mandatory case codes
Apex’s own log is the proof this does not work.
What Flock said
“In July 2025, we introduced an optional case code requirement for searches … we will now require case codes for all law enforcement searches by the end of the year. Agencies may bypass the requirement in emergencies.”

What Apex’s records show

The reason field existed for the entire period covered by Apex PD’s search log. An independent audit of that log, published through public records by HaveIBeenFlocked.com, counts at least 15,335 searches between July 29, 2025 and June 29, 2026. Twenty-eight percent of them recorded no reason at all. Blank is the single most common entry in the log. That is roughly 4,300 searches of a national tracking network with nothing written down.

So this was never a missing feature. It was an available field that went unused thousands of times, with no apparent consequence. Making the same box mandatory does not make the entry in it true, and Flock wrote an emergency bypass into the requirement. The deadline is also “by the end of the year,” which means as of today nothing has changed.

Verdict: Flock is fixing a blank field. Apex’s problem is 4,300 blank fields that nobody in authority ever asked about. Mandatory theater is still theater.
Claim 6 · Audit Assistance and proactive lockout
A four-month-old tool, adopted by a minority of customers, never independently evaluated.
What Flock said
“16 weeks ago, we introduced Audit Assistance, which detects abnormal activity and flags it for Administrator review … More than one-third of our customers have voluntarily adopted Audit Assistance. We will now require all law enforcement customers to adopt this feature by the end of this year.”

Read the numbers in that sentence

  • The tool is about four months old.
  • More than one third adopted it voluntarily, which means close to two thirds did not.
  • It becomes mandatory at the end of the year, not now.
  • It has never been evaluated by anyone outside Flock.

The denominator problem

Flock’s evidence that the tool works is that identification of misuse went up, and that recent officer arrests followed reviews. Up from what? As the ACLU puts it, without knowing how many officers are actually misusing the system you cannot tell whether the tool catches 95 percent of them or 5 percent. Arrests prove abuse exists. They do not measure detection.

The scale of the problem it is supposed to solve

These figures come from other people’s investigations and we cite them as theirs. The Washington Post reported on August 2 that at least 50 officers have been charged with or accused of misusing Flock and similar systems, and that in 26 of those cases investigators or prosecutors said the officers were tracking wives, girlfriends, exes, an ex’s new partner, or women they wanted to meet. One woman learned her officer ex had looked her up roughly 600 times. The Institute for Justice has published a public tracker of more than 100 documented ALPR abuse incidents, which it updates as new cases surface. Flock announced its changes eleven days after that Post investigation.

EFF makes the structural point better than we can: enhanced audit tools address a problem that the product itself created.

Verdict: Flock is selling the audit tool as the answer to abuse it has not shown the tool catches, and it will not be required anywhere until December.
Claim 7 · Accuracy
“Fewer than 9 human-reported errors per 1 million alerts” is a self-graded exam.
What Flock said
“This combined system of safeguards results in fewer than 9 human-reported errors for every 1 million alerts … an alert is a lead, not a conclusion, which should be confirmed by an officer.”

What that statistic can and cannot tell you

Human-reported is the operative phrase. The numerator counts only errors somebody took the trouble to report to Flock. A wrongful stop that ends at the roadside with an apology and no complaint never enters it. The denominator is alerts, not stops, not searches, and not people pulled over. Flock has not published the underlying method, and the independent security review it commissioned from Bishop Fox is not due out until September.

The framing also quietly moves the risk. If an alert is only a lead, then every consequence of a bad read lands on the officer at the car window and the driver inside it, not on the vendor that generated it.

What Apex has never published

A false positive rate. Apex PD told Council it had roughly 1,700 hotlist hits and offered ten example cases with no incident numbers and no arrest reports. Nobody outside the department can check what fraction of those hits were wrong.

Verdict: An unaudited accuracy number published by the vendor being audited is a marketing claim until an independent evaluator says otherwise.

What the announcement does not touch

The national sharing network still exists, unchanged
No warrant requirement to search historical location data
No resident vote anywhere in the process
Nothing in it is binding on Flock or on Apex
Apex still has no ALPR ordinance
Apex’s own retention is grandfathered at 30 days
No published false positive rate, here or nationally
Every change is reversible by the company at will

EFF asked the question that ends this argument: what is stopping Flock from reversing course on any of this if its law enforcement customers start defecting to another vendor? Nothing. That is what separates a setting from a law.

The Apex numbers this announcement does not change

Everything above is a reading of somebody else’s press release. The numbers below are ours, from Apex Police Department’s own records obtained under public records law. They are the reason this page exists.

15,335
Searches logged in 11 months, July 29, 2025 to June 29, 2026
6,896
Outside networks reachable by a single Apex search
100,075
Cameras reached at the high end of one search
28%
Searches with no reason recorded, the most common entry

Thirty-one Apex operators ran those searches, averaging about 61 per active day, with Sunday the single busiest day of the week and activity recorded in the 2 a.m. hour. Sharing snapshots show 994 organizations with active access in March 2026 and 1,070 in April. The full log analysis, including the source documents, is on our audit log page. The presentation Apex PD actually brought to Council is broken down in our June 9 analysis.

A vendor cannot give back consent it never asked for. Only a vote can do that.

Questions for the Town of Apex

These are answerable in writing, and the answers are either yes with a date or no.

What we are asking for

Every camera in Apex removed
ALPR banned in the Town Code, not in a policy memo
The ban covers any vendor, not just Flock
No searching other agencies’ ALPR networks either
Plate data already collected destroyed and certified
A recorded public vote, on the record

We will be straight about one thing, because it matters and because someone will point it out. Our ask did move. For months we pushed a governance ordinance: delete in ten days, a 99 percent accuracy floor, no mass sharing, a civilian committee with real authority. We built it in good faith on the premise that this system could be made safe if the conditions were strict enough.

We do not believe that anymore, and the reason is on this page. A vendor that can rewrite its own safeguards in a blog post is not a vendor you can bind with conditions. A department that left 28 percent of its searches unexplained was not short of rules, it was short of consequences. Conditions require a system that respects them.

So the ask is now removal and a ban. That is a bigger ask than we started with, and August 11 is why we think it is the right one. When the chamber overflowed, nobody in that room asked for a better-tuned dragnet.

Sources, in full

Every quotation attributed to Flock on this page is taken verbatim from Flock’s own August 13 post, linked below. Where a fact comes from someone else’s reporting or analysis, we say whose it is.

What we are not claiming

We have not seen how Evidence Mode actually operates, because Flock has not published its scope. We are raising the ACLU’s question about it, not asserting the answer.

We have not confirmed whether Apex PD has adopted the 7-day setting, enabled offense filtering, or turned on Audit Assistance. Those are open questions on this page for exactly that reason.

The national figures on Flock’s size, the June 2025 search-tool history, the officer misuse counts, and the abuse tracker are the work of the ACLU, EFF, the Washington Post, the Register-Guard, and the Institute for Justice. We cite them as theirs. The Apex search log, sharing snapshots, and contract records are ours.

If a fact on this page is wrong, tell us which one and where, and we will correct it and say that we did. Email contact@deflockapex.org.