OpenAI说它未能阻止一名加拿大大狙击手使用一个被禁止的ChatGPT账户,促成新的安全升级。 OpenAI says it failed to stop a Canadian mass shooter who used a banned ChatGPT account, prompting new safety upgrades.

OpenAI披露, 2026年2月10日杀害八人的大规模狙击手B. C. Tumbill Ridge的Jesse Van Rootselaar使用第二个ChattGPT账户, OpenAI revealed that Jesse Van Rootselaar, the Tumbler Ridge, B.C., mass shooter who killed eight people on February 10, 2026, used a second ChatGPT account after her first was banned in June 2025 for posts about gun violence. 该公司说,在她的身份得到加拿大皇家骑警的公开确认之后,它才发现第二个账户,此后,它改进了安全措施,包括更好地发现逃逸企图,与心理健康专家加强风险评估,以及与加拿大执法部门直接沟通。 The company said it only discovered the second account after her identity was publicly confirmed by the RCMP and that it has since improved safety measures, including better detection of evasion attempts, enhanced risk assessment with mental health experts, and a direct communication line with Canadian law enforcement. OpenAI说,它本来会在几个月前制定的一项新的安全协议中标注她的身份,但在最初的禁令期间,该系统并不活跃。 OpenAI stated it would have flagged her under a new safety protocol developed months prior, but the system wasn’t active during the initial ban. 加拿大官员批评该公司拖延采取行动,而OpenAI则强调不断努力加强保障措施,防止滥用其平台。 Canadian officials criticized the company for delayed action, while OpenAI emphasized ongoing efforts to strengthen safeguards and prevent misuse of its platform.