A hacker exploited Anthropic's Claude chatbot to steal 150GB of sensitive data from various Mexican government agencies between December 2025 and January 2026. By masquerading as a participant in a "bug bounty" program, the attacker tricked Claude into generating thousands of hacking plans, detailing exact targets and credentials. Curtis Simpson, chief strategy officer at Gambit Security, noted this incident highlights how easily AI can be weaponized, exacerbating vulnerabilities in government systems.