COINNEWSStartup Cathedral hacked Anthropic's systems using AI
Cathedral, a startup founded by former D.O.G.E. project employees, successfully breached Anthropic's internal systems. Researchers used artificial intelligence to find vulnerabilities in the infrastructure of the Claude model developer.
The incident caused serious concern in the White House. A US administration representative noted that if such a hack is accessible to a small group of young entrepreneurs, then similar actions could be carried out by China as well.
Access to confidential employee data
Anthropic confirmed the fact of unauthorized access to internal data. However, the company clarified that access was limited in terms of privileges.
Protected assets were not compromised
The research was conducted as part of Anthropic's official vulnerability search program (bug bounty). The company guaranteed that the source code, model weights, and user correspondence on the Claude service remained safe.
2K views
