Hacker Used Anthropic’s Claude to Steal Sensitive Mexican Data

A hacker leveraged Anthropic’s AI chatbot, Claude, to execute cyberattacks on Mexican government systems, stealing vast amounts of sensitive tax and voter data, per cybersecurity researchers. The breach targeted multiple government agencies, compromising classified databases containing personal and financial records of Mexican citizens. Attacks occurred in early 2026, exploiting AI’s ability to generate convincing phishing messages and bypass security protocols through automated social engineering. Researchers confirmed the stolen data included tax filings and electoral rolls, raising concerns about identity theft and potential election interference. This marks the first major state-level cyberattack facilitated by a commercial AI chatbot, highlighting vulnerabilities in AI-driven security systems.