Detroit Man's Arrest Highlights Racial Bias In Facial Recognition Tech

Our Take

Scripps News does a solid public service revisiting Robert Williams' case, a Detroit man arrested in front of his own kids because a facial recognition algorithm decided his face was close enough to a shoplifting suspect's. Close enough, it turns out, is not a legal standard — it's a coin flip dressed up in the language of science. The research is not new or contested at this point: face recognition systems misidentify Black and Asian faces at dramatically higher rates than white faces, and police departments have repeatedly treated those flawed matches as probable cause instead of a starting hunch.

This matters to us because facial recognition and automated license plate readers are cousins in the same surveillance family. Both sell departments on the promise of instant, computer-verified certainty. Both get deployed with minimal oversight, minimal audit trails, and maximum faith that the machine is right. Flock Safety's ALPR network, now blanketing thousands of towns, runs on the same premise — trust the algorithm, act on the alert — and it inherits the same risk of false positives snowballing into wrongful stops, searches, or worse, just with license plates standing in for faces.

Williams eventually got an apology and a settlement, but most people misidentified by opaque surveillance tech don't get either, because they never find out the algorithm was involved at all. If you want to see how much of this camera infrastructure has already landed in your own neighborhood, check our map, and if you'd rather your city not be the next case study, our take-action page has concrete steps for pushing back before the cameras go up, not after someone's in handcuffs.