Washington:A cybersecurity startup named Cathedral managed to access sensitive internal data at the artificial intelligence company Anthropic in less than 12 hours, as reported by The Washington Post.

According to Anadolu Agency – English, the months-old startup, founded by former members of the US Department of Government Efficiency (DOGE), identified vulnerabilities in third-party business software used by Anthropic, which is known for creating the Claude chatbot. The startup's researchers utilized AI agents to penetrate Anthropic's internal systems, accessing meeting transcripts, salary and job offer details, and Slack messages related to advanced AI models.

The cybersecurity testing was part of a program allowing external researchers to inspect Anthropic's systems for potential vulnerabilities. Anthropic confirmed that Cathedral had accessed a 'limited set' of its sales, human resources, and IT systems. The company stated that it had since closed the exposed pathways and informed federal officials about the breach.

An Anthropic spokesperson noted that Cathedral initially attempted to breach the company's own software before successfully exploiting weaknesses in third-party business tools. A senior Trump administration official indicated that the report on the incident has circulated within the government, raising concerns over how quickly the startup accessed sensitive systems. The official cautioned that the findings might suggest potential exploitation by foreign intelligence agencies.