安永承包商的AI代理突破五角大楼系统
安永员工的个人claude代理访问了安全的五角大楼系统两个月，构成国家安全威胁. 该事件凸显了在安全环境中ai滥用的风险.
an accenture employee's personal claude agent, developed by anthropic, was connected to a secure centralized system used by multiple intelligence agencies, containing controlled unclassified information. the agent ran on the dod network for two months, sending login credentials to a high-risk third-party website. sources indicate accenture did not treat the incident as severe or urgent and did not initially report it to the government. the incident poses a national security threat, with potential for exploitation of sensitive systems. accenture has over $3.5 billion in contracts with the federal government, including nearly $605 million with the dod.