Face Recognition is TERRIBLE, Cops Use It Anyways
The short version: NACDL's video lays out how faulty facial recognition still drives arrests, a warning sign for the same unchecked trust departments place in Flock's ALPR network.
Our Take
Credit to NACDLvideo for putting a spotlight on something defense attorneys have been screaming about for years: face recognition tech is routinely wrong, disproportionately wrong for Black and brown faces, and yet police departments keep leaning on it like it's gospel. The pattern the video describes — a flawed algorithmic match becoming the entire basis for probable cause, an arrest, a ruined week or a ruined year — should sound familiar to anyone tracking Flock Safety's ALPR network.
Facial recognition and license plate readers are cousins in the same surveillance family. Both are sold to city councils as neutral, infallible tools. Both get fed into investigations with almost no adversarial testing before a judge signs off. Both put the burden on the accused to prove the machine screwed up, rather than on police to prove it didn't. When a plate reader flags a car as 'associated' with a crime, or a face-match algorithm flags a suspect, the human in the loop too often just rubber-stamps the computer's guess instead of doing actual police work.
The video is a good reminder that 'the algorithm said so' is not evidence, it's a shortcut — and shortcuts in policing tend to land hardest on people who already have the least power to fight back. If you want to see how thick this surveillance net has grown in your own town, check our camera map, and if you're ready to push back on it locally, our take-action page has a place to start.
This is DeFlock The USA’s original commentary. The video above is the work of NACDLvideo, published on YouTube — full credit to the creator.