Home / Technology / AI Startup Breaches Top AI Firm Anthropic in Hours
AI Startup Breaches Top AI Firm Anthropic in Hours
26 Oct
Summary
- AI agents infiltrated Anthropic's systems within 12 hours.
- Sensitive data like salaries and transcripts were accessed.
- National security concerns raised over potential abuse by hackers.
A cybersecurity startup named Cathedral has reported breaching the systems of leading AI company Anthropic in less than 12 hours. The startup, formed by former members of the Department of Government Efficiency, stated that its AI agents exploited vulnerabilities in third-party business software to gain access.
Cathedral's draft report, reviewed by the Washington Post, indicated that sensitive information, including meeting transcripts, salaries, and Slack messages, was accessible. Anthropic acknowledged that researchers accessed a limited set of sales, HR, and IT systems, but asserted that core user data and AI models remained secure.
Anthropic stated the breach occurred during a cybersecurity testing program but accused Cathedral of rule violations. The incident has reportedly triggered national security concerns in Washington, with fears that such exploits could be used by state-sponsored hackers, particularly from China.
Cathedral, valued at $1.4 billion, focuses on military applications. The breach highlights broader concerns about the security of enterprise software used by even advanced organizations. This event contrasts with recent reports of AI agents being used to facilitate hacking, as Anthropic's systems were the target this time.
