MIT科技评论经过15个月的调查发现，美国边境耗资数十亿美元的AI监控系统存在严重失效，移民在摄像头监视范围内的死亡事件频繁发生1。该调查通过映射近4000具遗骸位置与近600个监控塔的数据，确认自2022年以来至少有138例移民遗体在监控塔名义覆盖范围内被发现，其中超过一半很可能在监控视野内1。
调查揭示的系统性缺陷包括破损的塔、检测算法失效和特工未能响应警报等问题1。以2024年4月跨越美墨边境的José Morales Bernal为例，他距离最近的监控塔仅360英尺，但最终死于"环境暴露"1。尽管存在这些问题，美国政府计划再投入10亿美元来扩大虚拟墙系统规模1。
MIT Technology Review conducted a 15-month investigation into the U.S. border's artificial intelligence surveillance infrastructure and found widespread system failures that have coincided with migrant deaths in monitored areas 1. By mapping the locations of nearly 4,000 remains against data from approximately 600 surveillance towers, researchers identified at least 138 cases since 2022 in which migrant remains were discovered within the nominal coverage zones of monitoring stations, with over half likely positioned within actual camera view 1.
The investigation documented specific failures undermining the system's effectiveness. José Morales Bernal crossed the U.S.-Mexico border in April 2024 within 360 feet of the nearest monitoring tower but ultimately died from environmental exposure 1. Systemic deficiencies identified in the report include damaged surveillance towers, malfunctioning detection algorithms, and Border Patrol agents failing to respond to alerts 1. Despite these documented problems, the U.S. government plans to invest an additional $1 billion to expand the virtual wall system 1.
评论
还没有评论，欢迎留下第一条。