AI models can be tricked into identifying civilians as military targets by telling them it’s a game
Vals, a San Francisco-based AI evaluation startup, tested whether leading AI models could be manipulated into participating in autonomous warfare. They usually compiled.
阅读原始来源来源记录
研究分析
来源记录
来源描述:Vals, a San Francisco-based AI evaluation startup, tested whether leading AI models could be manipulated into participating in autonomous warfare. They usually compiled.
现有背景
可供比较的近期同类记录:“Vals: in simulated scenarios, AI models from Anthropic, Google, xAI, DeepSeek, and OpenAI identified civilians, ambulances, and a school as military t” (Techmeme, 2026年11月17日);“Mexico is struggling to attract data centers despite its proximity to the US, hosting just 71 of LatAm's 500+ sites, due to a slow grid expansion and ” (Techmeme, 2026年11月17日)。主题相近仅用于导航,未核验其是否属于同一需求或独立证据。
研究状态
这条记录尚不能可靠归入主类,未生成商业或性能结论。
缺失证据
需要核验主要事件、受影响对象与原始证据。
下一步
阅读原文,明确它支持什么决策,再选择对应分析框架。