到2025年,圣保罗的AI监控系统逮捕了近4,000人,其中有许多是错误的,引发了对隐私和偏见的关切。 São Paulo’s AI surveillance system arrested nearly 4,000 people by 2025, including many wrongly, sparking privacy and bias concerns.

圣保罗Smart Sampa AI监视系统于2024年启动,在40,000个摄像头上使用面部识别仪识别逃犯和嫌疑人,声称抓获了3,000名逃犯,逮捕了近4,000人,导致到2025年抢劫事件减少近15%。 São Paulo’s Smart Sampa AI surveillance system, launched in 2024, uses facial recognition across 40,000 cameras to identify fugitives and suspects, claiming to have captured 3,000 fugitives and arrested nearly 4,000 people, contributing to a nearly 15% drop in robberies by 2025. 然而,超过8%的被捕者后来因错误而被释放,包括一名80岁的男子和精神病患者被错误地拘留。 However, over 8% of those arrested were later released due to errors, including an 80-year-old man and psychiatric patients wrongly detained. 至少141起逮捕涉及过时的逮捕令,许多目标个人面临民事罪行,如无偿儿童抚养费,引起人们对该制度用于民事执法的担忧。 At least 141 arrests involved outdated warrants, and many targeted individuals faced civil offenses like unpaid child support, raising concerns about the system’s use for civil enforcement. 批评者强调缺乏种族数据,使得算法上的偏差无法评估,并指出该系统对较贫穷的周边邻里和移民的影响格外严重。 Critics highlight a lack of racial data, making algorithmic bias impossible to assess, and note the system disproportionately affects poorer, peripheral neighborhoods and migrants. 虽然官员否认偏见,强调公共安全,民权倡导者却警告侵犯隐私和系统性不平等。 While officials deny bias and emphasize public safety, civil rights advocates warn of privacy violations and systemic inequities.