Exclusive: Trump tells "The Axios Show" that Anthropic was a national security threat
TL;DR
Trump told Axios he briefly saw Anthropic as a national security threat the previous week. After the G7 summit he softened the tone, saying Dario Amodei seemed 'nice' and 'smart' and that Anthropic responded quickly. Axios says an Amazon report about a vulnerability alarmed the administration. The US then restricted access to Anthropic's most advanced models for countries outside the US and for foreign nationals inside the US.
Nauti's Take
Trump is turning model risk into a loyalty and access fight. That is understandable if a frontier model can endanger critical infrastructure, but dangerous when interviews, rival warnings and personal impressions start steering government action.
Anthropic likes to present itself as the safety-first lab. Now it has to show that safety is an auditable operating standard, not brand positioning.
Briefingshow
The story shows how fast AI safety can move from technical dispute to state power in the US. For labs like Anthropic, better models and safety papers are no longer enough; they have to prove who gets access, how vulnerabilities are measured and when governments should trust their controls. Export controls can hit product plans and non-US customers directly.
Sources