FACES 人脸识别导致错误逮捕——但这并非首次
诉讼详细说明了运行已久的 FACES 系统生成的 93% 匹配度如何导致罗伯特·迪隆被错误逮捕,凸显了在警务工作中将算法相似度视为确定性的危险。
the aclu has filed a lawsuit against florida police departments after the FACES facial recognition system—a database containing tens of millions of images—generated a 93% similarity score that led to the wrongful arrest of robert dillon. despite dillon living 300 miles from the crime scene and having no prior connection to the area, officers treated the algorithm's match as a near-certain identification and allegedly failed to include exonerating evidence in warrant applications. the complaint highlights a fundamental misunderstanding of probability in policing, where a similarity score on facial features was conflated with the likelihood of identity. this incident serves as a critical warning for law enforcement agencies relying on legacy biometric tech, revealing how automation bias can override investigative due process and lead to severe civil rights violations.