据五名知情人士透露，OpenAI员工多年来一直就运营安全松散和内部威胁向领导层和董事会发出警告。
OpenAI Staff Raised Security Concerns for Years Amid Rivalry and Espionage Risks, Report Finds OpenAI employees have warned leadership and the board of directors for years about lax operational security and insider threats, according to five people familiar with the matter.
员工警告称，广泛的代码访问权限、未经审查的内部测试沙箱以及广泛的代码库共享，使关键的人工智能技术面临外国对手的威胁。
Workers warned that widespread code access, unvetted internal testing sandboxes, and broad repository sharing left critical AI technology vulnerable to foreign adversaries.
在10月13日前OpenAI工程师阿卜杜勒拉赫曼·阿尔福赞被捕后，担忧情绪进一步升级，他被联邦特工指控在高管会议室安装录音设备。
Concerns escalated following the October 13 arrest of former OpenAI engineer Abdulrahman Alfozan, who was charged by federal agents with planting a recording device in an executive conference room.
OpenAI回应称，公司高度重视安全，已将生产工程师重新分配以加强基础设施，并已隔离敏感模型权重并限制访问权限。
OpenAI responded that it takes security seriously, has reassigned production engineers to bolster infrastructure, and has isolated sensitive model weights and restricted access.