Anthropic 助力 NSA 网络作战，同时因伦理风险与五角大楼争执
anthropic 正在向 nsa 部署其 mythos 模型，以针对中国和伊朗的进攻性网络任务。这种合作与其公开的安全立场相矛盾，并标志着向激进军事 ai 采用的转变。
anthropic is actively supporting the nsa by placing engineers on-site to tailor its mythos ai model for offensive cyber warfare, specifically aimed at networks in china and iran. this deep collaboration appears to contradict anthropic's ongoing legal dispute with the department of defense, which previously blacklisted the company for refusing to enable mass surveillance and autonomous weaponry. the nsa partnership bypassed that ban, suggesting a differentiated policy where defensive us interests are prioritized. this move highlights the growing pressure on ai labs to support national security objectives, even at the expense of stated ethical guidelines.