Export control reaches the model itself, not the chips
On 12 June 2026 the US government issued an export control directive barring access to Claude Fable 5 and Mythos 5 by any foreign national, inside or outside the United States, including Anthropic's own foreign-national employees. Unable to verify nationality in real time, the company switched both models off for every user.
Why it matters
For years American export control addressed the chips and the tools models are trained with. Here it reached the model itself: what was restricted was not hardware but access by people of a given nationality to an already deployed system. Anthropic publicly disagreed with the basis, saying the government had supplied only verbal evidence of a narrow, non-universal bypass of the model's safeguards.
The directive reached the company at 5:21pm Eastern on 12 June 2026. The government cited national security authorities and gave no details of the concern itself. Anthropic's understanding was that it concerned a method of bypassing Fable 5's safeguards that would allow it to find software vulnerabilities. The company reviewed a demonstration of the technique: it surfaced a small number of previously known, minor vulnerabilities which, on the company's own check, other publicly available models including GPT-5.5 find without any bypass at all. Fable 5 and Mythos 5 shipped on 9 June and share one underlying model. Fable 5 was released with strong safeguards for general use; Mythos 5, with fewer, went only to a small number of trusted Project Glasswing partners for defensive cybersecurity. Because the order took effect immediately and there was no reliable way to verify nationality in real time, Anthropic disabled both models entirely. Its other models were unaffected. The company said it disagreed that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. Reuters supplies context the statement does not: the order arrived as an earlier dispute between the administration and Anthropic was easing. Earlier that year the company had refused to let the US military use its models for domestic surveillance and fully autonomous weapons, and the government responded by placing it on a supply-chain blacklist.