Artificial Intelligence Has a Racial Bias.Grainy photo mistakenly identified Blackman as shoplifter

Our Take

Dee Smith Lets Discuss lays out the Robert Williams case plainly: a grainy still from a shoplifting incident got run through facial recognition, spit out a bad match, and cops showed up on a man's front lawn to arrest him in front of his kids. No lineup, no real investigative work, just a machine's guess treated as probable cause. This wasn't some hypothetical bias study — it's a real person's life getting upended because a department trusted a black box over basic police work.

This case matters far beyond one algorithm vendor or one department. It's the same pattern we track with Flock Safety's ALPR network: automated systems get deployed with minimal oversight, treated as infallible by the officers using them, and the burden falls hardest on people who already get over-policed. Facial recognition and license plate readers are different tech, but they share the same failure mode — a computer flags you, and suddenly you're the suspect, due process be damned. Departments keep buying these tools faster than they're building in accountability for when the tools are wrong.

If you want to see how much of this automated surveillance infrastructure has already gone up in your area, check our camera map. And if you think your city council should be asking harder questions before signing more contracts with Flock or facial recognition vendors, our take-action page has steps you can use right now.