Trump says Anthropic no longer a US security threat

TL;DR:

  • Donald Trump told Axios he no longer regards Anthropic as a national security threat, walking back a stance he held “a week ago”.
  • The comments follow the company disabling worldwide access to its Fable 5 and Mythos 5 models after a White House order targeting foreign nationals.
  • Trump kept open the option of using Defense Production Act powers, leaving the truce short of a formal settlement.

President Trump has signalled an end to the fortnight-long standoff with Anthropic, telling Axios that while he might have seen the AI company as a national security risk recently, he no longer does. The remarks cap a dispute that forced the firm to pull access to its most capable models and drew the UK government into seeking an exemption.

A rapid de-escalation

The row began when the administration ordered Anthropic to block foreign nationals from its Fable 5 and Mythos 5 systems, prompting the company to disable access for everyone rather than police users individually. Trump credited chief executive Dario Amodei with responding “very quickly” and “responsibly”, and the two met alongside other G7 leaders at a summit in France. An Anthropic spokesperson thanked the administration for working to resolve the matter.

For UK readers, the climbdown matters because Britain was caught directly in the crossfire. The Prime Minister had pressed for a UK carve-out from the ban, a request the White House initially rejected, even as some early users retained access through existing arrangements. The episode exposed how exposed UK firms are to decisions made in Washington over the AI tools they have built workflows around.

Looking forward

The truce is informal rather than binding. Trump pointedly declined to rule out invoking the Defense Production Act, telling Axios “I have the power to use a lot of things” while adding he was not sure he would need to. That ambiguity leaves the underlying question unresolved: a single executive decision can still sever access to frontier models overnight. The saga has already hardened the case for open-source alternatives among UK organisations wary of building critical processes on infrastructure they do not control.