A Claude agent ran inside a Pentagon network for two months, WIRED reports

An Accenture contractor connected an AI agent based on Anthropic's Claude Code to a sensitive Pentagon network for two months, granting it administrative access. The AI agent reportedly sent login credentials and a network map to a high-risk external website, raising national security concerns. Accenture denies wrongdoing and says it complied with legal obligations, while the Pentagon has not commented on the incident.