Voices38 is back in action: a tool for cracking Denuvo in 2024 games is ready.

Hacker Used Claude and ChatGPT Chatbots to Hack Mexican Government

A hacker used Anthropic's Claude chatbot to carry out a massive cyberattack on Mexican government institutions. According to Bloomberg, 150 gigabytes of confidential information were stolen as a result. The attack lasted about a month, starting in December of last year.

As reported by specialists from Gambit Security, the attacker automated the hacking process using a neural network. Claude was used to search for vulnerabilities in government systems and generate scripts to exploit them. The chatbot created thousands of detailed reports with ready-made action plans, specifying which resources to attack next and which credentials to use.

Although Claude's security mechanisms initially blocked illegal requests, the hacker managed to bypass the restrictions using special prompts. Anthropic confirmed the incident, blocked the associated accounts, and conducted an internal investigation. The company assured that the latest Claude Opus 4.6 model already contains tools to prevent such abuses.

To enhance the attack, the perpetrator also used ChatGPT. With its help, he analyzed movement within the compromised networks, searched for ways to cover his tracks, and identified accounts with access to critical systems. OpenAI confirmed attempts to violate usage rules but noted that their systems did not assist the hacker.

The identity of the perpetrator has not yet been established. This case is not the first instance of Claude being used for criminal purposes. A year ago, Chinese hackers used the same chatbot for attacks on global systems. The recurrence of such incidents calls into question the effectiveness of security mechanisms even for the most advanced neural networks and underscores the need to strengthen cybersecurity measures in the era of active AI development.