- Florida man files suit alleging police relied on a faulty facial-recognition match.
Police arrested Robert Dillon in August 2024 after a system flagged him as a 93% match to a suspect captured on a McDonald’s camera. The match came from a low‑quality snapshot of a computer screen, not a clear image of the suspect. Officers ignored a license‑plate reader check that showed Dillon was over 300 miles away from the crime scene.
The case highlights how law‑enforcement can treat algorithmic output as proof, even when the data quality is poor and contrary evidence exists. If courts accept the suit, agencies may need stricter checks before using facial‑recognition as the basis for arrests.
Dillon’s lawyers say the police built a narrative around the algorithm instead of testing it, a pattern that could affect other jurisdictions relying on similar AI tools.