Inside the fight over Claude Mythos 5
TL;DR
Anthropic reportedly received a US export-control directive on Friday evening ordering it to block Mythos 5 and Fable 5 for all foreign nationals, including affected Anthropic employees. Because that was not cleanly enforceable, Anthropic disabled the heavily promoted models entirely and sent safety and cybersecurity staff to Washington for talks with the Trump administration.
Nauti's Take
This looks less like a mature safety process and more like improvised industrial policy under pressure. If one jailbreak report can freeze a flagship model worldwide, the bar is not yet clear enough.
Anthropic also helped create the trap with its own risk-heavy marketing: when a lab sells a model as almost too dangerous, regulators may eventually take that framing literally.
Briefingshow
The fight shows how quickly AI safety messaging can rebound on the company using it. Anthropic framed Mythos as unusually powerful, and the government is now using that risk narrative as leverage. For customers, model quality is no longer the only issue; political shutdown risk becomes part of the buying decision.
Sources