Arrested for being Black: When AI surveillance goes wrong | More in Common

Our Take

Localish's report on Michael Oliver's ordeal is a gut-punch case study in what happens when departments treat AI-driven identification as gospel. A misfire flagged him as a wanted felon, and instead of that flag being treated as a lead to verify, it became probable cause for a traffic stop and arrest. This is the exact failure mode we've warned about with Flock Safety's network: automated systems generate alerts with false confidence, and officers act on them as if a machine can't be wrong.

The racial dimension here isn't incidental. Facial recognition and pattern-matching systems have documented higher error rates on Black faces, and when you stack that on top of ALPR networks that already over-surveil certain neighborhoods, you get a pipeline where technology doesn't just fail neutrally, it fails in ways that land hardest on people who are already over-policed. Flock's cameras aren't doing facial recognition, but they're part of the same broader trend, letting algorithms make consequential decisions with minimal human skepticism baked into the process.

Cases like this are why transparency and pushback matter before the cameras go up, not after someone's already been cuffed on a sidewalk for a mistake they had nothing to do with. Check our camera map to see what's already watching your area, and if your city is considering a new surveillance contract, our take-action page has tools to get involved before it's too late.