Build1 distinct publisher3 min readPublished
Alpharetta's entire sharing policy is one radius setting that auto-approves agencies within 500 miles, and the lieutenant who runs it could not say when it was switched on. That gap points to a logging problem sitting underneath the policy question.
The Engineer · Build desk

Compiled by The EngineerSomething wrong?How this is made
A geographic auto-approve rule is a standing grant evaluated against the requester, not against the case. Alpharetta approves law-enforcement requests from inside 500 miles automatically, and agencies outside that ring have to ask individually, Lieutenant Jason Hiott told WIRED [8]. Draw the circle: pi times 500 squared is roughly 785,000 square miles of standing eligibility, set by one control in an admin panel [1]. The check does not look at the offense or the date range; it looks at where the requester sits.
Then the graph. Taking the three reported counts at face value, 2,000 outbound relationships plus 1,300 inbound, measured against 2,300 distinct nodes, puts about 1,000 organizations on both sides of Alpharetta's network [2]. Reciprocity is the product. Flock's value rises as customers connect deployments, because a plate read in one jurisdiction lets an investigator follow a vehicle into the next county or state [15]. Every edge accepted for that reason is also an edge somebody else reads from.
The detail worth putting on the review slide is that Hiott could not say how long the 500-mile setting had been active, or whether the department had ever changed Flock's default configuration [9], a bigger gap than the 2,000 figure itself. A permission system without a durable, queryable change history cannot answer the question that arrives with the first complaint: who was eligible to see this plate on this date. Flock says sharing is off by default for new and onboarding customers, and would not say whether that had always been the default [10]. The current default therefore tells you nothing about the configuration an older account is still running.
Treat 2,000 the way you would treat someone else's benchmark table. It is a property of Alpharetta's settings and of how many Flock customers sit within 500 miles of north Georgia. For the number to transfer to your city, your radius scope and your neighbour density both have to resemble theirs. It also counts permission rather than queries: WIRED did not establish how many outside organizations searched records captured in Alpharetta, how often, or what investigations prompted any search [5]. Converting a permission count into an exposure count needs per-requester query logs, and none were reported. Access itself arrives in three shapes, automatic under an administrator's settings, individually approved, or a standing right to ask [6], and one headline figure flattens all three.
Which is where the purchasing conversation goes. Flock's pricing page publishes no standard price, and quotes vary by sensor count, technology, software package and coverage area [14]. Hardware was the cheap part from the beginning: Langley found existing readers priced at about $25,000 per camera and prototyped with an Android phone in a waterproof box [12]. What a city buys now is a permissions engine with cameras attached, so the line items worth arguing over are radius scope, video restrictions, request limits and log retention, all of which the admin console already exposes [7]. Runtimewire's reading is that access controls, offense filters and audit systems are core infrastructure here rather than compliance add-ons [16], and the unanswerable question about the 500-mile setting is the evidence for it.
Ranked by verification strength, evidence, and original report placement.
Hiott could not say how long the 500-mile setting had been active or whether Alpharetta had changed Flock's default configuration.
Runtimewire's analysis: Alpharetta shows how several dozen local cameras can create permissions for thousands of outside organizations, making access controls, offense filters and audit systems core infrastructure rather than compliance add-ons.
Data from several dozen Flock Safety cameras in Alpharetta, Georgia is accessible to more than 2,000 outside organizations, WIRED reported on August 28.
In the other direction, Alpharetta receives Flock data from more than 1,300 entities.
WIRED counted more than 2,300 distinct nodes across Alpharetta's incoming and outgoing sharing relationships, after accounting for organizations appearing on both sides.
The Alpharetta cameras connect the department to police agencies, colleges, airports, prosecutors, drug task forces and government investigators around the country.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 29, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
build
Flock's OS Investigate starts the search before there is a suspect1 distinct publisher
invest
Flock's $8.3B asset is a searchable national archive. Its liability is the same archive1 distinct publisher
product
Flock's sharing settings plugged one Atlanta suburb's plate reads into 2,000 organizations1 distinct publisher
product
Alpharetta counted 85 Flock searches against two officers only after it opened an investigation1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Documentary, but relayed through a single outlet
The underlying material is unusually hard for a surveillance story — public-records counts, 549 pages of Reynoldsburg audit logs, a named lieutenant on the record, an on-the-record vendor statement about defaults. None of it reaches us directly: Runtimewire is summarising WIRED, and no second newsroom has touched the same records. The soft spot is the deployment itself, still described only as 'several dozen' cameras with no number or placement, while the 2,300-node figure is stated to three significant digits.
Permissions proven, searching mostly not
This is a live, dense network rather than a pilot: 2,300-plus distinct counterparties around one suburb, receiving-side growth of about 42 percent and sending-side growth of about 125 percent since June 2025, and an Ohio log showing a single officer sweeping 5,392 networks in one sitting. What stays unmeasured is the specific thing the headline implies — how often anyone actually pulled Alpharetta's records, which the reporting openly says it could not establish.
Number invites more than it says
'2,000 organizations' counts permissions, and Runtimewire says so in its fourth paragraph — that discipline is why the gap is small rather than zero. The strain comes from what travels: a figure built from auto-approval settings reads like 2,000 agencies watching a Georgia suburb, and the piece's own headline does nothing to slow that reading. On the other side, one detail is arguably underplayed rather than overplayed — a vendor that cannot reconstruct when a nationwide-search feature was enabled is a bigger finding than the node count, and it appears deep in the story.
The moat is the setting under scrutiny
Flock earns more as the graph densifies — the reporting states outright that value grows when customers interconnect — so the company's commercial interest points toward permissive sharing, while its public position is that customers own their data and decide access. That allocation is convenient: it puts a 500-mile radius on a lieutenant who cannot say when it was set. Note also which statements here are self-descriptions. The 'off by default' claim comes from the vendor, without any account of past defaults; the 'no live feeds or analytics' assurance comes from the department, uncorroborated by logs.
Trustworthy relay, thin corroboration
The factual spine is consistent and attributable: named officer, named company statements, records-based counts, and a clear statement of what was not established. Our confidence is held back by structure rather than sloppiness — one publisher, itself downstream of the original investigation, with the camera count vague, the derived figures (785,000 square miles, roughly 1,000 overlapping organizations, about 19 nodes per officer) resting on floors rather than exact counts, and the truncated closing section leaving the competitive comparison unfinished.