UK opens phase three of medical AI regulation sandbox
TechNews reports that the UK government adopted all 44 recommendations from an NHS doctors-led medical AI commission and opened phase three of the MHRA's AI Airlock sandbox. The article says this establishes lifecycle-based monitoring of medical AI in real clinical environments, including ongoing performance tracking and safety assessments after major model updates. It also examines how responsibility may be divided between clinicians and technology developers when AI-assisted diagnostics contribute to patient harm.
The reported policy shift moves medical AI oversight beyond one-time approval toward continuous monitoring, while the accompanying legal analysis highlights accountability for both clinicians and developers.
人工智慧正以驚人的速度跨入極度強調安全與精準的醫療核心。從 X 光與 CT 放射影像的自動判讀、癌症早期篩檢，到複雜病理診斷，AI 診斷系統已成為許多醫院不可或缺的輔助工具。
然而，當演算法出現誤判，導致醫師做出錯誤的醫療處置或延誤患者黃金治療期時，這筆沉重的醫療傷害責任，究竟該算在站在第一線的醫師頭上，還是開發演算法的科技公司身上？
英國政府宣布全盤採納由國家醫療服務體系（NHS）醫師領軍的醫療AI監管委員會提出的44項建議，並由英國藥物及保健產品管理局（MHRA）開啟AI Airlock沙盒計畫第三階段，正式啟動全生命週期的醫療AI動態監管機制。
#UK government#NHS#MHRA#AI Airlock#NHS doctors-led medical AI commission