The US put AI under export control: Anthropic switched two models off worldwide
Washington barred foreign nationals from Fable 5 and Mythos 5. Rather than fence access by nationality, Anthropic turned both off for everyone — three days after launch.
On 12 June 2026 the US government ordered Anthropic to suspend its Fable 5 and Mythos 5 models for all foreign nationals, citing an export-control directive grounded in national security. The order came just three days after the models launched.
Rather than attempt to fence access by nationality across a global product, the company switched both models off entirely, for everyone, worldwide. It stands as one of the first cases in which AI models were handled like a conventional export-controlled item.
Anthropic did not slow its release cadence through the episode. On 24 July 2026 it announced Claude Opus 5, offering a 1M-token context window, up to 128K output tokens, adaptive thinking on by default and a five-level effort setting. Opus 5 was the fourth model the company shipped in under two months.
The next step the industry expects is Fable 5.1. Claims circulated in late July said the model was finished internally and due in August at unchanged pricing — $10/$50 per million tokens — with its release deliberately timed against OpenAI's next model.
MHA Analysis
This is one of the first thresholds at which AI models were handled like a conventional export-controlled good, and its precedent value is high. That Anthropic switched the models off for everyone rather than filtering by nationality shows how hard it is, technically, to gate a global product by citizenship. For non-US markets — Türkiye and Central Asia among them — the implication is blunt: access to frontier models is now a geopolitical variable rather than a commercial decision, which turns domestic model development from a prestige project into a strategic one.
