Ex-DOGE Staffers Hack Anthropic in 12-Hour AI Agent Attack

Remember DOGE? Elon Musk’s Department of Government Efficiency, which was appointed by Trump to cut costs and streamline federal operations. Its former employees are now hacking AI giant Anthropic.

Using AI agents, four ex-DOGE employees breached Anthropic's internal systems in 12 hours. The incident came to light the same day Anthropic filed for its initial public offering.

How Did a Small Startup Get Inside Anthropic?

DOGE, the cost-cutting unit run by Musk and Vivek Ramaswamy for the Trump administration, implemented cuts in 2025 that led to over 250,000 federal workers losing their jobs.

DOGE shut down on July 4, 2026, with minimal attention. Within three weeks, four former employees emerged with a $1.4 billion military cybersecurity startup called Cathedral.

Cathedral was founded by Gavin Kliger, Luke Farritor, Marko Elez, and Jack Stein.

According to Anthropic, the researchers spent days trying to breach its own software but failed. The AI agents then targeted third-party tools used for sales and administrative tasks.

Through those routes, they accessed meeting transcripts, salary data, job offers, and Slack conversations about advanced AI models, a Cathedral draft report viewed by the Washington Post stated.

“The fact that weaknesses in ordinary enterprise software can compromise the infrastructure of even the most advanced organizations should be a wake-up call for every company in America,” the Post indicated, citing Cathedral’s draft report.

The hack was not illegal; it was carried out under Anthropic's bug bounty program, which rewards external researchers for discovering and reporting vulnerabilities.

BREAKING: AI cybersecurity startup founded by ex-DOGE staffers breaches Anthropic's internal systems in under 12 hours using AI agents, accessing sensitive employee data & Slack messages. — Polymarket (@Polymarket) October 26, 2026

Why Is Anthropic Pushing Back?

Anthropic acknowledged that Cathedral accessed a 'limited set' of its sales, HR, and IT systems. The company stated that customer chats, source code, and model weights—the files that operate Claude—remained untouched.

“Our monitoring flagged some of Cathedral’s activity before they reported it to us, and we shut off their access the same day,” the report stated, citing Anthropic.

Anthropic further claimed that Cathedral violated program rules by failing to disclose all its activity and exaggerated the national security threat.

In July, Cathedral raised $160 million at a $1.4 billion valuation, with Andreessen Horowitz and Sequoia Capital leading the round. A senior Trump administration official told the Washington Post that three federal agencies have engaged the startup.

Anthropic is moving forward with its Nasdaq IPO after a public filing on Monday. Nigel Green, CEO of deVere Group, described its reported valuation of up to $2 trillion as 'nuts'.

Anthropic reported that it found no evidence that any other party exploited the vulnerabilities.