The Cameras Aren't the Problem. The Network Is.

Flock Safety spent the last two years becoming the most recognizable name in automated license plate reading, and most of the last twelve months defending it. Cities have deactivated or cancelled contracts at a steady clip since early 2025, with counts ranging from roughly thirty to over eighty depending on who is doing the counting and how they define a cancellation. Flagstaff, Santa Cruz, Fort Collins, Oshkosh, Columbia Heights. Leon County, Florida pulled $440,000 in grant funding. Senator Ron Wyden wrote the FTC. The ACLU ran a campaign with a name built for a headline.

It is worth understanding what the objection actually is, because the answer determines whether it has anything to do with a construction site.

What people are objecting to

The Flock backlash is not a referendum on cameras. It is a referendum on two specific design choices.

The first is the shared network. Flock operates something in the range of 80,000 to 100,000 cameras across thousands of jurisdictions, and the value proposition to a police department is that its cameras are not only its own. A search initiated in one town can reach across state lines into networks owned by agencies that never agreed to that specific query. In May 2025, 404 Media reported that a Johnson County, Texas sheriff's search touched 83,345 cameras across 6,809 networks in a single query. Reporting that same year documented thousands of lookups run on behalf of federal immigration enforcement through informal access rather than formal process. Flock has since added administrative controls and ended certain federal pilots. The damage to the brand was already done.

The second is that a government agency is the user. The Fourth Amendment restrains government searches, and the live legal question, argued in cases like Schmidt v. City of Norfolk, is whether a dense, persistent, warrantless camera network constitutes a search at all. A district court said no in January 2026. It is on appeal to the Fourth Circuit. That question is genuinely unsettled, and it deserves to be litigated carefully.

Neither of those conditions exists on a fenced jobsite.

The private-property distinction is not a technicality

A general contractor recording its own active site is not a state actor conducting a search. It is a property owner documenting its own premises. In Louisiana, Mississippi, Alabama, Texas, and Florida, video-only recording on commercial property is broadly permissible without consent, excluding areas carrying a reasonable expectation of privacy such as restrooms and changing areas. All five are one-party-consent states for audio, though audio raises additional considerations worth handling deliberately rather than by default. Signage is sound practice. It is not, in most commercial video contexts, a statutory precondition.

The footage stays with the owner. It does not enter a searchable multi-jurisdictional index. Nobody in another state runs a query against it. The entire structural complaint driving the Flock cancellations is absent, because the architecture is different.

That distinction has been lost in a news cycle that treats "surveillance" as a single undifferentiated thing. It is not. Ownership of the data, and who may query it, is the whole argument.

Which brings us to what the cameras are actually for

The construction industry has historically bought site cameras for deterrence. Put up a tower, add a strobe and a speaker, and thieves go somewhere else. Deterrence is real and it is measurable, and equipment theft justifies the spend on its own. Industry estimates put annual construction equipment theft losses somewhere between $300 million and $1 billion, with recovery rates well under a quarter, though the underlying data traces to a mid-2010s reporting series that has not been updated with the same granularity since. Copper theft costs the grid roughly a billion a year by Department of Energy estimate.

Deterrence is the easy argument. It is also the smaller one.

The larger value of site video is documentation. A slip-and-fall claim filed four months after the fact. A subcontractor disputing sequence of work on a delay claim. A workers' compensation filing whose described mechanism of injury does not match what the camera saw. A subrogation opportunity that exists only if you can establish which party's equipment caused the loss. In each of those, the question is not whether someone was deterred. It is whether anyone can prove what happened.

Insurers require preservation of evidence. Surveillance systems overwrite on a loop. The gap between those two facts is where liability exposure lives, and it closes only if footage can be located and pulled before the loop erases it and before a preservation letter turns into a spoliation problem.

Retrieval speed is therefore not a convenience feature. A system that takes two weeks to produce an incident clip is a system that will occasionally produce nothing. Sites that run on grid power and fixed infrastructure have an additional problem, which is that the highest-risk periods on a construction project are the ones before permanent power, permanent fencing, and permanent anything. Solar towers exist because the exposure starts on day one and the infrastructure does not.

Purpose-driven infrastructure

Treating cameras as a security line item puts them in competition with guards, fencing, and lighting, where they will be evaluated on whether anything got stolen last quarter. Treating them as documentation infrastructure puts them alongside the daily report, the safety log, and the certified payroll, where the value shows up months later in a claim file.

The Flock controversy is a public-sector governance problem about who may search whose data. It belongs to the civil liberties conversation, and it should be resolved there. A contractor deciding how to document its own site is answering a different question entirely: when something goes wrong on your property, how long does it take you to prove what happened, and who else has a copy.

Those are separable questions. The news cycle has not been treating them that way, and the confusion costs contractors clarity they need.