Cathedral hacked Anthropic's infrastructure using AI agents

Cybersecurity startup Cathedral used autonomous algorithms to gain unauthorized access to the internal networks of Anthropic. The incident occurred as part of a bug bounty program.

During the attack, the specialists utilized third-party software through which the agents infiltrated the corporate infrastructure. Representatives of the developer of Claude systems officially confirmed the data compromise.

The investigation revealed access to information from the sales, HR, and IT departments. However, the company's management disputes the depth of the penetration into internal secure perimeters.

The event demonstrated the vulnerability of leading AI labs to threats related to third-party software dependencies in the software supply chain.