Anthropic Employees Accuse Trump Administration of Targeting Them
TL;DR
Anthropic employees accuse the Trump administration of singling out the company after Washington limited access to its newest Claude models, Fable 5 and Mythos 5, through export-control action. The order reportedly covers foreign nationals inside the US as well as Anthropic employees. Anthropic responded by pulling access more broadly instead of trying to block only selected user groups. The administration cites national security and possible jailbreak risks, but has not clearly explained the legal basis in public.
Nauti's Take
Anthropic is not automatically the neutral hero here: the company benefits when its models are framed as unusually powerful and security-critical. At the same time, an opaque government move against one specific vendor looks like policy by instinct.
That combination is toxic: PR inflates the danger, the state overreacts, and users lose trust in core AI infrastructure.
Briefingshow
This is bigger than two models going offline. If the US governs frontier AI through one-off orders instead of clear rules, vendors, customers and international partners cannot plan around stable access. For Europe, it is a warning that dependence on closed US models is also a political operational risk.
Sources