Cathedral Researchers Access Anthropic Systems in Under 12 Hours

Researchers at cybersecurity startup Cathedral accessed a limited set of Anthropic’s internal business systems after running AI agents for less than 12 hours, in a test conducted through Anthropic’s security research program. Anthropic says it closed the exposed pathways and found no evidence that anyone else exploited them. The reported findings describe access to sales, human resources and IT systems after researchers exploited vulnerabilities in business software used for administrative and sales work. Cathedral’s report said the researchers found paths to meeting transcripts, salary and job-offer details, and Slack messages about Anthropic’s advanced models.