The meeting
Anthropic CEO Dario Amodei walked into the White House on Friday for a meeting with Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent. The subject: Anthropic’s new Mythos AI model and whether the company and the Trump administration can find a path forward.
Why it matters: Anthropic is the maker of Claude, one of the most capable AI systems in the world, and the US government simultaneously wants to use it and punish the company that built it. How this dispute resolves will shape AI governance for a generation.
The Pentagon’s position
The Defence Department designated Anthropic a “supply chain risk” earlier this year, a label previously reserved for companies linked to foreign adversaries such as Huawei and Kaspersky.
The Pentagon’s argument is straightforward: it needs unrestricted access to Claude for all lawful purposes, especially in wartime. With US forces engaged in the Iran conflict, military planners say AI tools are essential for logistics, intelligence analysis, and operational planning.
Anthropic’s position
Anthropic has said it wants to work with the military but draws a line at autonomous weapons and mass surveillance. The company argues that AI models are not yet reliable enough for life-or-death decisions without human oversight, and that US law has not been updated to govern these uses.
The company sued the administration over the blacklisting. A federal judge in California blocked the government from extending the supply chain risk label beyond the Defence Department. But the DC Circuit Court of Appeals ruled the Pentagon can maintain its own ban while the legal challenge plays out.
The reality on the ground
A RedState investigation published on Saturday found that multiple federal agencies outside the Pentagon are quietly using Claude despite the official blacklist. The California court ruling that limited the ban’s scope appears to have created a grey zone that agencies are exploiting.
President Trump told reporters he had “no idea” Amodei was meeting White House officials, a claim that drew scepticism given that Wiles is his most senior adviser.
What happens next
The courts will determine whether the Pentagon’s supply chain risk designation can survive legal challenge. In the meantime, the White House meeting suggests back-channel diplomacy is under way.
The broader question is whether the US can maintain AI leadership while simultaneously blacklisting the company behind one of its most advanced models. Supporters of the administration say national security requires full government access. Critics argue that punishing companies for setting safety boundaries will drive AI development offshore.
Neither side has shown willingness to compromise on the core issue: who decides when AI is ready for the battlefield.