Anthropic’s Mythos mess is only getting worse
TL;DR
Anthropic took Mythos 5 and Fable 5 offline for two weeks after an export-control order from the Trump administration. The rule blocked access for foreign nationals, including people inside Anthropic. After tense negotiations, Mythos 5 can return for a small set of approved cybersecurity organizations and infrastructure providers. Fable 5, the public-facing model in the same class, is still stuck with no clear rollout timeline.
Nauti's Take
This is not a normal launch delay, it is release management by government mood swing. If you build on frontier models, you now need exit plans, fallback models, and contract language for sudden access cuts.
Model choice is officially infrastructure risk.
Briefingshow
This is bigger than Anthropic. The US is effectively testing a new form of AI export control before it has a clean process for applying it. For labs, customers, and global partners, the message is blunt: even released frontier models can be politically frozen if they are treated as a security risk.
Sources