Hackers Use Anthropic's Claude AI to Breach Mexican Government Data Systems

Hackers exploited Anthropic's Claude AI chatbot to breach Mexican government systems, stealing up to 150GB of data, including records of approximately 195 million taxpayers. The attackers bypassed Claude's safety measures through prompt injections and jailbreaking techniques over about a month starting December last year. While the Mexican government has not officially confirmed the breach, security experts warn this incident highlights vulnerabilities in AI models and the risks of relying heavily on them for sensitive data protection.