Before buying a single camera: how Milton, West Virginia, is locking down AI surveillance by law

By Zendoric · October 10, 2026.

The city council of Milton, a small town in West Virginia, gave first approval on Tuesday to an ordinance on artificial intelligence surveillance. It does not regulate a technology the city already uses: it regulates one it does not have. That is the peculiarity of the case, and what makes it relevant far beyond its few thousand residents.

The rule, which will have its second and final reading on November 10, neither authorizes nor bans automated surveillance. It establishes the procedure the city would have to follow if it ever wanted to adopt it: a public hearing before seeking funding, acquiring a system or signing a contract with a vendor; a public "use policy" that the council must expressly approve; an annual results report open to residents; and the obligation to cancel, as soon as legally possible, any contract signed before the ordinance that fails to comply with it. The text defines the surveillance it covers fairly broadly: automated license plate readers, biometrics, gunshot detection, location tracking and cameras. Patrol car cameras, emergency recordings, utility meters and police body cameras are excluded.

The reason given by Mayor Shane Evans is simple, and fairly representative of the whole debate: according to the article, he does not believe the technology is yet ready to genuinely protect the community's rights, and while he acknowledges it can be positive in some respects, he worries that the information will end up being used for other purposes. In August, the same council had already voted in favor of creating an ordinance against the use of Flock Safety — the network of AI license plate readers used by thousands of police departments in the United States — or any equivalent AI surveillance camera. The current ordinance turns that one-off rejection into a permanent procedure.

Broadly speaking, Flock Safety's name does not appear here by chance: in recent years, various journalistic investigations in the United States have documented cases of local police departments that, through the national cross-camera search network the company offers, allowed license plate queries by federal immigration agencies or by police from other states, without a warrant and, in many cases, without the affected residents knowing. That track record explains why a small council, with no incident of its own pushing it, decides to act before buying anything.

Our reading: what is interesting about Milton is not the content of the ordinance — similar to those dozens of US cities have already passed over the past decade under the banner of so-called community control over police surveillance — but the order in which it arrives. The usual sequence with this kind of technology is: a vendor runs a demonstration, the police adopt it with minimal administrative approval, it is deployed, and only when an abuse or a data leak emerges does policy react after the fact. Milton reverses that sequence: it builds the authorization, the transparency and the public report before having anything to watch. It is preventive governance in its most literal form, and it fits something we have been observing: the brake that is really starting to contain AI in the United States is not a specific federal law on artificial intelligence — which still does not exist in consolidated form — but the ordinary exercise of powers a city hall already had (approving contracts, requiring hearings, publishing reports), now applied to a new technology.

In the short term, this has a real cost, and it should not be glossed over: every municipality that takes this path delays, or outright rules out, tools that, well governed, could help public safety, from automatic gunshot detection to faster location of a stolen vehicle. The distrust Mayor Evans expresses is not gratuitous alarmism: it is the reasonable response to an industry that has yet to settle, in a way convincing to the average citizen, who has access to the data it generates, for how long and for what purpose. As long as that data governance problem remains open, it is understandable that small cities prefer to wait.

In the long run, however, the pattern Milton represents is good news, not bad. The more municipalities build frameworks of transparency and accountability before adopting algorithmic surveillance, the easier it will be for the next generation of these tools — with biometrics and analytical capabilities far beyond today's — to be deployed with safeguards built in by design, instead of chasing abuse after it has already happened. The abundance AI can bring does not depend only on models getting better; it also depends on the institutions adopting them learning, as this council in a West Virginia town is doing, to say "not yet" without saying "never."