Facial Recognition: Innocent Arrests Due to AI Errors #shorts
The short version: A short from Dave Linthicum Is Not AI revisits how facial recognition misfires have landed innocent people in handcuffs, a cautionary tale that applies just as much to license plate AI.
Our Take
This short is a quick, blunt reminder of something we say constantly: AI confidence is not the same as AI accuracy. Facial recognition has already put innocent people in jail cells because an algorithm said "match" and a human being didn't push back hard enough. Credit to Dave Linthicum for keeping this in front of casual viewers who might otherwise assume face-scanning tech is as reliable as a fingerprint.
Swap "face" for "license plate" and you've described the core problem with Flock Safety's ALPR network. These cameras feed probabilistic guesses — plate reads, "similar vehicle" alerts, vague descriptions like color or missing plate covers — into police databases, and officers treat those guesses as leads worth a stop or worse. The same brittle pattern-matching that misidentifies a face can misidentify a plate, and there's rarely a transparent audit trail showing how often that happens or who it happens to.
The fix isn't trusting the vendor's marketing, it's visibility and friction: knowing where these cameras actually are and giving communities a real say before they go up. Check our camera map to see what's watching near you, and if you want to push back on unaccountable deployments, our take-action page has next steps.
This is DeFlock The USA’s original commentary. The video above is the work of Dave Linthicum Is Not AI, published on YouTube — full credit to the creator.