Salem Fired Its License-Plate Cloud and Kept the Cameras, Because Data Is the Real Haunted House

Salem has done something unusually radical for a city that can make a zoning meeting feel like a constitutional convention: it read the terms and conditions.

The North Shore city is moving away from Flock Safety’s automatic license-plate reader network over concerns about how vehicle-location data is stored and shared, while keeping the basic camera function for investigations and missing-person cases. The latest update, reported by The Salem News and GovTech on July 27, says the city’s replacement cameras are already installed at high-traffic locations and should be operating by Aug. 31. Salem is still deciding what software will run them.

That last sentence is the whole story. The camera is the visible object: a small box staring at traffic while everyone pretends not to notice. The consequential technology is the database behind it, the access rules around it, and the answer to a question municipalities have historically treated like a terms-of-service footnote: who owns the record of where a car went?

Salem Wants the Useful Part Without the National Data Buffet

Automatic license-plate readers, or ALPRs, photograph passing vehicles, extract plate numbers, attach time and location, and make those records searchable. In practical terms, that lets investigators look for a vehicle connected to a robbery, find a missing person’s car, or reconstruct movements after an incident. It is not magic. It is a very fast index built from a lot of ordinary people driving to work, the grocery store, a doctor’s appointment, or the one coffee shop in Salem that has somehow achieved a 45-minute line.

Flock’s pitch is that a cloud platform makes the system more useful. A department can search its own cameras, connect to a larger network, and use software that recognizes vehicle characteristics beyond the plate itself. The network effect is the product: more cameras and more participating agencies create more searchable context.

That is also where the privacy problem lives. Salem’s official explanation says the city found “insufficient protections” for data generated by Flock’s systems and that the data should be owned and used only for Salem’s public-safety priorities. The city is not rejecting surveillance as a category. It is rejecting the idea that the most sensitive part of the system should sit inside a vendor’s broader network with rules that local residents did not write.

That distinction matters. A city can decide that a targeted camera used under clear rules is a worthwhile investigative tool. It can also decide that a private company’s ability to determine how data is accessed, shared, retained, or improved is a dealbreaker. Those are not contradictory positions. They are the difference between buying equipment and outsourcing governance.

The Cloud Is Where the Camera Gets Its Personality

The technical architecture is not especially mysterious. Cameras collect images. Optical-character-recognition software turns plates into text. A cloud service stores the records, applies search and alert tools, and controls permissions. The software can be genuinely useful: a detective does not want to watch six hours of traffic video by hand when a searchable event log can narrow the field to three vehicles.

But the cloud layer also decides what “local control” means. Can an officer search only Salem’s records, or can the query travel across a national network? Is a record deleted after a short retention period, or held for weeks? Does the city receive the raw data, or merely a vendor-shaped view of it? Can the vendor continue using the information to improve its service after the contract ends?

The ACLU has warned that Flock’s 2026 terms added a perpetual license allowing the company to use customer data to support and improve its services, along with language giving the company substantial control over the method and timing of access. Flock disputes the caricature. In a July 27 response, the company said customers own and control their data from collection through deletion, with controls for access, sharing, and retention.

That disagreement is not a side argument for lawyers to enjoy in a conference room. It is the product. When a surveillance company sells “public safety,” it is selling a set of technical defaults and legal permissions as much as it is selling cameras. Salem’s decision is an attempt to move those defaults back into city hall, where at least residents know which building to yell at.

City-Owned Hardware Is Not a Privacy Force Field

Here is the part where we resist the easy victory lap. Replacing Flock with city-owned cameras does not automatically make the system private, accountable, or wise. It simply changes who has to do the homework.

Salem will need a clear retention schedule, documented search rules, audit logs, access controls, security updates, public reporting, and a process for handling requests from outside agencies. It will need to decide whether its software can perform vehicle-type searches, how alerts are reviewed, and whether police can query the system without a case number or other justification. It will need to budget for maintenance and cybersecurity instead of treating the cameras like municipal furniture that becomes free after installation.

This is the governance equivalent of bringing a server rack home from the cloud. The invoice may look simpler. The responsibility is not.

The city also has to preserve the legitimate uses that made the cameras attractive in the first place. Salem says the devices have assisted major investigations, including cases involving shootings, robberies, sexual assaults, and the bombing of the Satanic Temple. A replacement system that is so locked down it cannot help solve crimes is a policy failure. A system that is so open it creates a searchable map of residents’ lives is another kind of failure.

Massachusetts Has Been Running This Experiment in Public

Salem is not an isolated cranky municipality. The ACLU of Massachusetts says more than 80 police departments in the state have signed contracts for Flock’s ALPR technology. That gives the Salem decision an audience well beyond Essex County. Every Massachusetts town considering a camera network now has a live case study in what happens when the hardware works but the data relationship stops feeling acceptable.

It also fits a larger Boston-area pattern in which the interesting technology is not the flashy device but the control plane behind it. Akamai’s Cambridge work on identifying and governing AI agents asks who gets access to the web. Liquid AI’s efficiency bet asks where intelligence should run and who pays for the compute. Walden Robotics is learning that factory robots need fleet management and safety systems, not just excellent demos. And the Massachusetts AI Coalition is trying to turn ecosystem enthusiasm into shared infrastructure.

Salem’s version is less glamorous and more important than most launch-stage technology. It is asking whether a public agency can use modern sensing without surrendering the rules to a vendor whose business model rewards more data, more connections, and longer memory. That question will follow every smart-city project from Boston to Beverly, usually in a PDF with a procurement number and a title like “integrated situational awareness.”

Verdict: A Useful Incremental Move With a Large Footnote

Salem has made a meaningful, limited move. It did not abolish license-plate readers. It did not pretend that privacy concerns make investigations disappear. It chose to keep the tool while taking more responsibility for the system around it.

That is promising, but the replacement software will decide whether this becomes a model or merely a municipal rebrand. City ownership is valuable only if it comes with enforceable rules, visible audits, short retention, narrow access, and enough technical competence to prevent “local control” from meaning “the same surveillance, now with a city password.”

Still, Salem deserves credit for identifying the real fault line. The future of public-safety technology will not be decided by whether a camera can read a plate. That part is already boringly possible. It will be decided by whether communities can keep useful systems without letting private platforms quietly turn public roads into a permanent behavioral archive.

My verdict is “useful incremental move,” with the usual Massachusetts requirement that everyone involved read another 200 pages before lunch. Salem kept the camera, fired the cloud, and gave the rest of the state a practical assignment: if the data is public power, the public should get to write the operating manual.