AI 实验室面临间谍渗透风险，安全专家称内鬼威胁已成现实
情报机构盯上模型权重这事儿被安全从业者亲口证实了。对比各家实验室还在卷外部防御的做法，内鬼审查是个被低估的缺口。做机密数据保护的团队和关心前沿模型安全的人都该看看她的判断。
AI 安全组织 SL5 Task Force CEO Lisa Thiergart 在 AI Deep Dive 节目中表示，间谍已实际进入 AI 公司内部工作，威胁并非假设。她指出前沿模型的安全防护不能只针对外部攻击者，必须同时防范内部人员。据她描述，掌握模型权重和训练细节的内部员工正是情报机构最想渗透的目标。Thiergart 称这一判断来自 SL5 Task Force 对行业安全状况的观察。这番表态把 AI 实验室的商业机密保护问题从理论层面推向了现实安全讨论。
当前主流 AI 公司的安全投入大多集中在网络防御与外部入侵检测，对内部人员审查和权限管控的讨论相对少。Thiergart 的观点提示前沿实验室需要把人事背景调查、最小权限设计纳入模型安全体系。对模型权重价值越高、员工流动越频繁的行业环境，这类内部风险越难通过常规技术手段消除。
The spy threat to AI labs isn't hypothetical anymore.
@SL5TaskForce CEO @LisaThiergart says spies are already working inside AI companies.
On this episode of AI Deep Dive, she explains why securing frontier models means accounting for insiders as well as outside attackers.
🚀 Watch the full episode of AI Deep Dive: https://t.co/UV75URyOTl
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI regulation, compliance frameworks, and sovereign compute policy. Key entities and technical topics under continuous tracking include AI安全, 前沿模型, 模型权重, SL5 Task Force.
Source attribution: curated from The Information (@theinformation).
