Anthropic’s Mythos mess is only getting worse
TL;DR
Anthropic has kept Mythos 5 and Fable 5 offline for two weeks after the Trump administration issued June 12 export-control demands over security concerns. Despite heavy talks in Washington, there is still no timeline: Anthropic says it has no news, and nobody knows whether or when the models will return. The fight hits a key business bet: Mythos was meant to lift revenue before a possible IPO, with input tokens reportedly priced at twice Opus 4.8.
Nauti's Take
This is not a clean safety process; it is a case study in policy uncertainty. When a model can be tested, launched, monetized, and then abruptly frozen through an unclear government process, the result is less safety and more operational chaos.
Anthropic does not look in control, but the bigger issue is in Washington: serious AI regulation needs clear thresholds, procedures, and authority. Otherwise every frontier release becomes a political bet.
Briefingshow
This shows how messy AI export control can become when politics, security, and product launches collide in real time. Instead of clear rules, the industry is seeing ad hoc decisions that can pull a flagship model from the market overnight. For US AI companies, that is a business risk; for customers, it is a warning that even premium models may not be reliably available.
Sources