Breaking The Creepy AI in Police Cameras

Our Take

Benn Jordan's deep dive into how Flock's AI actually works is the kind of technical scrutiny this industry desperately needs and rarely gets. Most coverage of Flock Safety stays at the surface level — a city council vote here, a privacy complaint there — but Jordan pulls back the curtain on the machine learning doing the heavy lifting: reading plates, tagging vehicle attributes, and quietly building a searchable history of everyone's movements. That's the part elected officials never seem to grasp when they approve these contracts.

What makes this worth your time isn't just the technical breakdown — it's the reminder that a private startup, not a government agency, built the infrastructure now logging your car's location nationwide, often with little transparency about how the underlying models are trained or what they're capable of inferring. Once that data exists in a database, it doesn't matter what a company's marketing page promises. It can be queried, shared, subpoenaed, or leaked. Jordan's willingness to actually poke at the technology rather than just repeat vendor talking points is exactly the kind of independent scrutiny that keeps companies honest.

Credit to Benn Jordan for putting in the work here. If this video has you wondering whether one of these cameras is already watching your street, check our camera map and see what you can do about it on our take-action page.