AI powered cameras are becoming a permanent fixture of city infrastructure. License plate readers, traffic cameras, and fixed surveillance systems are increasingly connected to AI that can flag vehicles, identify patterns, and alert authorities in real time. The capability is significant, and adoption is accelerating across thousands of municipalities.
A recent Fast Company piece looks closely at this expansion and raises a set of concerns that deserve attention. The article focuses on automatic license plate reader networks, where camera data is typically stored in the cloud, aggregated across jurisdictions, and shared between agencies and vendors under contracts that vary widely in how much oversight and transparency they include. In several documented cases, that data has ended up being used for purposes well outside its original scope, and city officials in some cases have said they weren’t fully aware of how widely their data was being shared.
The core issue isn’t the technology itself. It’s what happens to the data once a system goes live, and who actually controls that decision.
Capability and accountability are not the same thing
AI powered cameras can genuinely help public safety teams respond faster. Identifying patterns, surfacing relevant footage, and flagging anomalies in real time are valuable capabilities that legacy systems can’t match. That part of the equation is well established.
What the Fast Company piece highlights is that capability alone doesn’t guarantee good outcomes. A system that flags useful information is only as trustworthy as the structure around it. Who has access to the data? How long is it retained? Can it be repurposed for something other than what it was deployed for? Does the organization that owns the cameras actually understand and control how that data moves once it leaves the local system?
Without clear answers to those questions, even a well functioning detection system can create operational friction . Trust erodes quickly once people learn that data is circulating more widely, or being used for purposes other than what they were told.
What responsible deployment actually requires
As AI becomes more embedded in physical security and public infrastructure, the expectations placed on these systems are rising accordingly. Operational performance is no longer enough on its own. Organizations deploying AI powered cameras need to be able to answer, clearly and specifically, how their data is handled, who has access to it, and how decisions involving that data are made over time.
That means architecture matters as much as accuracy. Systems that keep data within the operator’s own control, rather than routing it through third party cloud infrastructure with loosely defined sharing terms, give organizations a real answer to those questions instead of a contractual promise.
This is where we’ve focused our efforts at Koshee. Building systems that deliver real time understanding while keeping control, transparency, and data ownership front and center. Koshee Protect processes video on site, within your own network, so the organization deploying the system retains full ownership over how that data is stored, accessed, and used. This provides clear visibility into where footage resides and how it is accessed, because it remains within your infrastructure.
When operational capability and data ownership are both built into the foundation, organizations get the full value of AI powered security: faster response, better situational awareness, and a level of trust that holds up over the long term, not just at the moment a system is deployed.
Read the full Fast Company article: Why AI-Powered City Cameras Are Sounding New Privacy Alarms
Learn how Koshee keeps your data under your control: protect.koshee.ai