Anthropic Employees Accuse Trump Administration of Targeting Them
TL;DR
The NYT reports internal Anthropic chats where employees ask whether the Trump administration is pressuring the company over 'bad vibes' and whether officials want Anthropic pushed out of the market. The trigger is a US order covering Mythos 5 and Fable 5. Anthropic took the models offline because foreign nationals, including its own US-based employees, were supposed to lose access.
Nauti's Take
Anthropic spent years warning that powerful models may need hard limits, and now it is getting the blunt version of state intervention. A 90-minute export-control ultimatum is not a serious governance model.
AI safety needs transparent evaluations, defined authority, and workable timelines. Otherwise the winner is the lab with the best White House access, not the one doing the strongest safety work.
Briefingshow
This shows how quickly US AI policy can move from rules to case-by-case power. When export controls, safety reviews, and personal access to officials blur together, customers and developers start pricing in political shutdown risk alongside technical risk. That is dangerous for US labs trying to sell globally trusted infrastructure.
Sources