Cathedral hacked Anthropic's infrastructure using agents

AI startup Cathedral has successfully attacked Anthropic's systems using specialized AI agents.

They managed to breach the perimeter in less than 12 hours. The attackers gained access to internal company messages and detailed personal information of employees.

Representatives of the targeted company stated that the leak affected only the internal environment: they claim that customer database information and the source code of the language models remained secure and were not compromised.

Such incidents demonstrate the growing risks of using autonomous systems for cyberattacks on the corporate infrastructure of tech giants.