Earlier this year, Anthropic refused to let the U.S. military use its AI models for domestic surveillance and fully autonomous weapons systems.
The government’s response was swift and harsh: the Defense Department labeled Anthropic a “supply-chain risk” in March; the first time that designation, normally reserved for companies tied to adversarial nations, was applied to a U.S. firm. It effectively bars tens of thousands of contractors from using Anthropic’s AI for military work.
Now, months later, both sides seem to be thawing – at least on the surface.
The relationship started improving after CEO Dario Amodei visited the White House in mid-April, the first face-to-face meeting since the dispute broke out. Anthropic employees also met with Treasury Secretary Scott Bessent to discuss Mythos, the company’s most advanced AI system, and potential presidential actions on AI.
Those conversations reportedly helped shape Trump’s executive order on June 2, which asked leading AI developers to submit their most advanced models for cybersecurity testing. Anthropic publicly said it looked forward to collaborating with the White House on the order.
Anthropic has also briefed National Cyber Director Sean Cairncross on Mythos and how to protect critical infrastructure from AI-enabled cyberattacks and warned that Mythos could actually supercharge such attacks.
But the Pentagon situation is a different story. Defense Secretary Pete Hegseth denied Anthropic’s request to reconsider the supply-chain risk designation, and both sides were still filing opposing legal briefs as of Thursday. Anthropic was also notably absent from an Army-run AI cyberattack simulation in April that included Google and OpenAI.
The backdrop to all of this: Anthropic is targeting a $1 trillion valuation for its IPO, and having a federal denylist designation hanging over you while trying to go public is not a great look.