Cathedral startup gained access to Anthropic data in under 12 hours — Anadolu Agency

US cybersecurity startup Cathedral gained access to part of the internal data of Anthropic, the developer of the Claude chatbot, in less than 12 hours during authorized testing of the company’s systems. Anadolu Agency reports this, citing The Washington Post, whose journalists reviewed a draft report.

Vulnerabilities in third-party services

Cathedral is a startup founded by former employees of the US Department of Government Efficiency (DOGE). Its researchers identified vulnerabilities in third-party business software used by Anthropic.

According to The Washington Post, the researchers used AI agents to find paths into the company’s internal systems. They gained access to meeting transcripts, salary information and job offers, as well as Slack messages discussing advanced artificial intelligence models.

More current news is available on the UA.News Telegram channel Telegram.

Anthropic’s response

The testing took place under a program that allows independent cybersecurity researchers to check Anthropic’s systems for vulnerabilities. The company confirmed that Cathedral accessed a “limited set” of its sales, human resources and information technology systems.

Anthropic said it had closed the identified access paths and informed federal officials about the incident. A company representative said that Cathedral spent several days trying to find a way to access Anthropic’s software before moving on to third-party business tools.

According to the publication, a senior Trump administration official said the report had been circulated within the government, and the speed of access to sensitive systems had raised concerns. He also warned that similar weaknesses could potentially be exploited by foreign intelligence services.