Read Max Talking Anthropic, Iran, and the Department of Defense with John Ganz
Mar 5, 2026
A lively talk about Anthropic’s $200M Pentagon contract and the company’s red lines on surveillance and lethal autonomy. They follow the leak-driven backlash involving Palantir and political attacks. Conversation shifts to how political tribes and narratives shape AI policy. They close by connecting global events and Elon’s Grok positioning to shifting priorities.
AI Snips
Chapters
Transcript
Episode notes
Anthropic Inserted Two Usage Red Lines
- The Pentagon contracted Anthropic for ClaudeGov but the company imposed two hard usage red lines: no mass domestic surveillance and no fully autonomous lethal targeting.
- Those clauses aimed to keep Claude out of legally and ethically fraught applications and were central to rising tensions between Anthropic and the DoD.
DoD Pushback Framed As Ideology And Control
- Pentagon figures, led publicly by Pete Hegseth, pushed back claiming contractors can't dictate legal limits and accused Claude of being 'too woke.'
- The dispute leaked into public speeches and briefings, revealing internal DoD factionalism about relying on Claude.
Palantir Call Sparked A Leak About Maduro
- An Anthropic exec asked Palantir on a routine call whether Claude was used in the Maduro capture, revealing curiosity about operational deployments.
- A Palantir exec then called the Pentagon to complain they sensed disapproval of Claude, escalating tensions via internal leaks.