- Published on
How a Contractor’s Claude Agent Got Into the Pentagon
- Authors
- Name
- DeepHorus
A defense contractor connected Anthropic's Claude Code to a secure Department of Defense system, WIRED reported this week. The setup gave a personal AI agent access to a Pentagon-linked environment. The disclosure surfaced on the Uncanny Valley podcast, where hosts Brian Barrett, Leah Feiger and Zoë Schiffer examined the case.
The hosts described the use as one of the most outlandish government AI deployments seen so far. The case raises immediate questions about how commercial AI tools are vetted inside sensitive networks. It also highlights the growing reach of agentic software beyond conventional corporate settings.
The Pentagon revelation arrived the same week Anthropic filed its S-1 for an initial public offering. Anthropic would be the first major US company focused solely on AI software to go public. It is also expected to raise more money than any company in history.
Anthropic's valuation climbed from $183 billion a little over a year ago to $965 billion in its most recent round. Some observers expect it to approach $2 trillion once trading begins. The filing showed billions in revenue alongside even larger losses and hundreds of billions committed to compute.
The filing is a stress test for the AI industry and the broader US economy. Its warnings about existential AI risk add unusual stakes for public investors. The Pentagon episode similarly shows commercial agents moving quickly into sensitive government systems.
On the podcast, the hosts framed the contractor case as a cautionary example of AI adoption outpacing oversight. They treated Anthropic's IPO paperwork as a mix of green, yellow and red flags. The episode also covered volunteers preparing for possible ICE activity at polling sites before the midterms. For Anthropic and government users, scrutiny will intensify as the IPO and Pentagon use draw attention.
