夜雨聆风
State documents reveal an AI drug tracking system failed to uncover repeated fentanyl theft by a nurse at Tennessee’s Erlanger Baroness Hospital. The medical worker stole leftover surgical fentanyl for months and showed clear intoxication symptoms at work until coworkers reported him. After failing a drug test, he was dismissed and confessed his misconduct. The hospital’s Sentri7 AI monitoring software, meant to flag missing controlled substances, ignored multiple abnormal drug records which should have set off alarms.
州政府文件披露，美国田纳西州厄兰格巴罗尼斯医院的 AI 药品监测系统未能发现一名护士长期盗取芬太尼。这名医护人员连续数月窃取术后剩余芬太尼，在岗时出现明显药物中毒症状，最终被同事检举。药检不合格后他被开除并坦白盗药事实。医院配备的 Sentri7 人工智能监测软件本应预警管制药缺失，却数次遗漏本该报警的异常用药记录。
The case sparks wide discussion over hidden drawbacks of popular hospital AI surveillance tools. Developed by a Dutch tech company, Sentri7 is used in hundreds of American medical centers, yet there exists no mandatory regulation to report its operational failures. Proprietary algorithm confidentiality blocks public inspection on system defects, making it hard to learn the real frequency of AI malfunctions across the industry. Experts hold conflicting opinions on whether the fault stemmed from software bugs or improper manual operation.
此事引发业界热议，主流医院 AI 监控系统暗藏短板。荷兰企业研发的 Sentri7 在全美数百家医院投入使用，但现行法规不强制上报系统故障。算法专利保密机制使得系统缺陷无法公开核查，行业内 AI 失灵的真实频次无从统计。业内专家对故障源于软件缺陷还是人为操作不当观点不一。
While AI medication monitoring has gradually replaced traditional manual inspection to curb illegal drug diversion, the accident proves automated technology is not flawless. Industry insiders warn that over-dependence on AI creates security gaps, especially inside operating rooms with flexible drug usage. A combination of intelligent monitoring and regular human spot-checks remains essential to secure hospital pharmaceutical safety.
尽管 AI 药品监管逐步替代人工核查以防范药品盗用，本次事故证实自动化技术并非万全之策。业内人士提醒，过度依托人工智能会滋生安全漏洞，用药流程灵活的手术室更是重灾区。智能系统搭配常态化人工抽查，才是保障院内药品安全的关键。
