70% Error Rate: Are Police AI Cameras Flagging Wrong Drivers?

Our Take

Isiah Factor: Uncensored is spotlighting something we've been saying for a while: these systems are marketed as precision tools, but the data tells a messier story. A 71% error rate on stolen-vehicle hits in a California city isn't a rounding error — it's a coin flip with worse odds. Every one of those false positives is a real driver getting pulled over, questioned, or worse, based on an algorithm's guess.

The deeper problem is accountability. When a human officer makes a bad call, there's at least a paper trail and a face attached to the decision. When an AI model flags a plate wrong seven times out of ten, who owns that mistake? Flock? The department that bought the subscription? Nobody, usually — and that's exactly why these numbers rarely surface unless a journalist or local reporter forces the issue, like this video does.

Error rates like this aren't just a tech glitch, they're a civil liberties problem hiding behind a dashboard. If a city near you is running Flock cameras, check our map to see what's already deployed, and if you want to push back on blind trust in these systems, our take-action page has concrete steps for getting answers from local officials.