Anthropic’s Mythos mess is only getting worse
TL;DR
Anthropic took Mythos 5 and Fable 5 offline after a June 12 export-control order from the Trump administration, affecting foreign staff as well as customers. After two weeks of talks, Mythos 5 can return only for a small approved group of cyber defenders and infrastructure providers. Fable 5 remains stuck with no clear rollout timeline, and the original directive is still in force; Washington says it can reassess the rules again.
Nauti's Take
The partial return of Mythos 5 sounds like progress, but it is mostly damage control. Fable 5 is still blocked, the core order still stands, and the precedent is now clear: Washington can treat AI models like export-controlled goods on short notice.
That uncertainty is bad for customers, developers, and investors. The PR-heavy talk of progress hides the real issue: nobody has a clean, repeatable process yet.
Briefingshow
This is bigger than Anthropic: if frontier models need case-by-case government approval, AI product planning becomes politically unstable. For companies, even paid enterprise access may suddenly depend on nationality rules, security labels, and government-approved customer lists.
Sources