A cybersecurity startup founded by former U.S. Department of Government Efficiency (DOGE) staffers claims its AI agents accessed Anthropic’s internal systems in under 12 hours. The company, described as a months-old startup called Cathedral, alleges it used AI agents to exploit vulnerabilities in third-party business software to reach internal data.
Across the reports, the focus is on what was reportedly accessed within Anthropic’s business environment, including information tied to staff and workplace communications. One outlet says the accessed material reportedly included meeting transcripts, salary details, job offers, and Slack messages discussing advanced AI models, while others summarize the claim more generally. None of the provided summaries attribute the claims to Anthropic or include verification details.
The coverage largely aligns on the same core allegation: Cathedral’s AI agents allegedly found a route into Anthropic’s sales, HR, and IT systems, doing so quickly. Differences mainly reflect how specifically outlets describe the alleged data and how narrowly they frame the incident as a “hack.”
