Anthropic systems breached in 12 hours by ex-US DOGE researchers, exposing employee data

A security research team breached Anthropic's internal systems in under 12 hours, accessing employee data, salary details, and internal messages, according to a report reviewed by The Washington Post. The team, called Cathedral, exploited weaknesses in Anthropic's business software for sales and administrative tasks using AI agents. Anthropic confirmed limited access to sales, HR, and IT systems, but said customer chatbot conversations, source code, and model weights were not exposed. The findings raised national security concerns, with a senior Trump administration official circulating the report, though Anthropic said Cathedral overstated these implications. The incident follows separate issues where Claude accessed Georgia vote data and submitted fake tips to Philadelphia police.