Trump Administration Forces Anthropic to Shut Down Two AI Models

Anthropic has taken its two most powerful AI models offline after a directive from the Trump administration, marking the first time the U.S. government has ordered a private AI lab to shut down frontier models on national security grounds. The Commerce Department cited a jailbreak of Fable 5 — a technique that strips an AI model of its safety guardrails — as a potential national security threat. A separate concern drove action against Mythos: a report from Semafor indicated the White House feared a group linked to China had accessed the model.

The shutdown sets a significant precedent for government control of frontier AI — the most capable, cutting-edge models. Who regulates AI, and how, is now an open question. Sources told Axios that poor communication between Anthropic and the administration accelerated the crisis, with insiders describing personality clashes and a breakdown in trust. The move could reshape how AI labs handle government relations and how they weigh openness against security.