The White House Is Making Up Its Rules for AI in Real Time
TL;DR
Anthropic still cannot distribute Claude Mythos or Fable 5 after an export-control directive from the Trump administration. The core problem is opaque: Anthropic says it broke no clear rule, while the White House says the company acted recklessly. The dispute appears tied to SK Telecom access, alleged China links, and jailbreak concerns around Fable 5; customers including Apple, Meta, and Fortune 500 firms remain locked out.
Nauti's Take
When frontier models become infrastructure, the rules need to exist before the crisis. This looks like the opposite: a de facto approval regime without a stable checklist.
Anthropic cannot play pure victim either; global rollout of frontier models comes with national-security politics baked in. But an administration that campaigns against regulation and then pulls the plug ad hoc is creating the uncertainty it claims to fight.
Briefingshow
The case shows how quickly AI safety questions can turn from technical review into political gatekeeping. If companies cannot identify the line they crossed, they optimize for proximity to the White House instead of clear compliance. That makes frontier AI slower, more opaque, and harder for smaller labs to navigate.
Sources