Startup Cathedral, founded by former DOGE staffers, claims to have breached Anthropic's internal systems

Cathedral, an AI cybersecurity startup founded by former DOGE staffers, claims in a draft report reviewed by The Washington Post that its AI agent used vulnerabilities in third-party commercial software to enter Anthropic's internal systems and, in under 12 hours, found access paths to sensitive information including meeting notes, salaries, offer letters and Slack messages, and that it notified Anthropic. Anthropic confirmed that researchers entered a “limited portion” of its sales, human resources and IT systems, including customer and partner business contact information and some basic user account records, but said Claude user conversations, source code and model weights were not accessed. The operation was part of Anthropic's security testing program that rewards vulnerability researchers, and Anthropic accused Cathedral of failing to disclose the full scope of its activity and violating program rules. According to a senior Trump administration official who spoke on condition of anonymity, the matter has raised national security concerns in Washington.