They Thought This Man Was a Criminal Based on His Face #wrongfularrest #surveillance

Our Take

Credit to Business Reform for putting a face and a name to what usually gets buried in a press release. Travis Williams isn't an abstraction or a statistic — he's a guy who got treated like a suspect because a piece of software said he looked like one. That's the quiet horror of algorithmic policing: the machine doesn't have to be right, it just has to be confident, and a lot of departments have decided that confidence is good enough to detain someone.

This isn't just an NYPD story. It's the same logic baked into the automated license plate reader networks we track every day. Flock cameras don't identify faces, but they do the same basic trick — feed a match to an officer and let human judgment get lazy downstream. Once a system tells a cop 'this is the guy' or 'this is the car,' the burden of proof quietly flips. Suddenly it's on the innocent person to prove the algorithm wrong, not on the state to prove it right. Wrongful stops, wrongful arrests, and ruined afternoons (at best) or ruined lives (at worst) are the predictable result of outsourcing suspicion to a black box.

The fix isn't more cameras or better PR from the vendors selling them — it's accountability and the ability for the public to see where this surveillance actually lives. Check our camera map to see what's watching your neighborhood, and hit take action if you want to push back on unchecked automated policing before it puts someone else in Travis Williams' position.