It was a meeting that would have been unthinkable two months ago: last Friday, Anthropic CEO Dario Amodei sat in the White House — across from Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent. Both sides called the conversation ‘productive and constructive.’ For Anthropic, it’s a small breakthrough after months in political exile.
The Backstory: Pentagon vs. Anthropic
Quick recap of how things got this far: in late February, Anthropic refused to give the Pentagon unrestricted access to Claude. The red lines: no autonomous weapons without human oversight, no domestic mass surveillance. The Pentagon then classified Anthropic as a ‘supply chain risk’ — a label normally reserved for companies with ties to foreign adversaries. Trump called Anthropic a ‘radical left woke company.’ Meanwhile, OpenAI signed its own military deal.
Why the Thaw Now?
The key is Mythos. Anthropic’s new AI model, unveiled in early April, achieved something no amount of lobbying could have: it got the government’s attention — for the right reason. Mythos is exceptionally good at finding security vulnerabilities in software. Under Project Glasswing, Amazon, Apple, Cisco, Microsoft, and others are already using it to secure critical infrastructure.
That makes Mythos too valuable for the government to ignore. According to an administration source, ‘every agency’ except the Department of Defense wants to use Anthropic’s technology.
What Was Discussed
Officially: cybersecurity, America’s lead in the AI race, and AI safety. Anthropic emphasized ‘shared interests.’ The White House spoke of ‘opportunities for collaboration’ and ‘shared protocols for challenges’ in scaling AI.
What wasn’t discussed — at least officially: the Pentagon dispute. Anthropic continues to call it a ‘narrow contracting dispute’ while simultaneously challenging the supply chain designation in court.
My Take
This is realpolitik in the AI age. Anthropic refused to play along with autonomous weapons — and is being punished for it. But at the same time, they’re building a model that strengthens national security. The White House faces a dilemma: you can’t simultaneously say ‘they’re a security risk’ and quietly want their cybersecurity tool.
Whether this thaw becomes a lasting peace remains to be seen. But the fact that the meeting happened at all shows something: in Washington, technology always wins over ideology in the end.
Sources: