The White House Is Making Up Its Rules for AI in Real Time
TL;DR
The Trump administration used an export-control directive to force Anthropic to take Claude Mythos and Fable 5 offline. Nearly a week later, the models still cannot be distributed. The dispute remains opaque: Anthropic says it did not break any concrete procedure, while the White House argues the company acted recklessly with frontier technology.
Nauti's Take
The White House cannot posture as anti-regulation and then block models through improvised pressure when a launch becomes uncomfortable. If Anthropic missed real national-security risks, the answer is a durable process with clear criteria, not retroactive power tests.
Otherwise the winners will not be the safest labs, but the ones with the best political early-warning system.
Briefingshow
The issue is not that the US wants to scrutinize high-risk AI models. The danger is the mix of anti-regulation messaging, unclear red lines, and immediate sanctions. If rules only appear after a model launch, labs will optimize for political signalling instead of predictable safety processes.
Sources