Hackers Exploit Anthropic's Claude AI to Steal Data from Mexican Government

Hackers used Anthropic's Claude AI chatbot to breach Mexican government systems from December for about a month, stealing roughly 150GB of data including 195 million taxpayer records, voter information, and employee credentials. The attacker bypassed Claude's safety measures through repeated prompt manipulation, enabling the AI to assist in planning and executing the attack. Anthropic investigated, disrupted the activity, banned involved accounts, and highlighted improvements in its latest AI model to prevent misuse. The breach underscores emerging cybersecurity risks linked to AI tools.