英国皇家警察督察署(HMICFRS)发布报告指出,面部识别技术的错误与算法偏见正加剧无辜者——尤其是少数族裔——被错误逮捕的风险 1。该机构重点审查了“回溯性面部识别”(RFR)技术的使用情况,发现目前缺乏充分的法定框架来规范该技术的应用 1。此外,调查人员常将RFR产生的结果误认为是确凿证据而非情报线索,导致在核实不足的情况下采取执法行动 1。
报告揭示,由于算法存在固有偏见,少数族裔更有可能被系统错误标记为嫌疑人 1。同时,上传低质量或模糊图像导致匹配结果不可靠的问题也普遍存在,且监管方发现本应删除的拘留图像仍被违规保留 1。针对这一现状,民权组织 Big Brother Watch 的代表 Jasleen Chaggar 批评警方未记录 RFR 的使用情况,并指责其利用带有偏见的设置搜索数百万非法持有的图像 1。对此,国家警察局长委员会负责人 Ben Russell 回应称,警方致力于以合乎道德、适度且负责任的方式利用该技术打击犯罪 1。
A report by His Majesty’s Inspectorate of Constabulary (HMICFRS) has warned that errors in facial recognition technology and algorithmic bias could lead to the wrongful arrest of innocent people, particularly those from ethnic minorities 1. The inspection focused on Retrospective Facial Recognition (RFR), finding that some investigators failed to distinguish between RFR results as intelligence leads versus concrete evidence, resulting in law enforcement actions taken without sufficient verification 1. The watchdog noted that algorithms are more likely to incorrectly flag minority groups as suspects, a risk exacerbated by the upload of low-quality or blurry images that produce unreliable matches 1.
Regulatory findings also highlighted significant compliance issues, including the retention of custody images that should have been deleted, alongside the absence of a comprehensive legal framework governing the technology's use 1. Jasleen Chaggar of the civil liberties group Big Brother Watch criticized police forces for failing to record their use of RFR and for searching millions of illegally held images using biased settings 1. In response to these findings, Ben Russell, head of the National Police Chiefs’ Council, emphasized that police remain committed to utilizing the technology in an ethical, proportionate, and responsible manner to combat crime 1.
评论
还没有评论,欢迎留下第一条。