What happened
On 12 June 2026, Commerce Secretary Howard Lutnick issued a directive under the Export Administration Regulations (EAR) suspending access to Anthropic's Fable 5 and Mythos 5 models by any foreign national, anywhere in the world — including foreign national employees inside the United States. To achieve compliance, Anthropic disabled both models globally for all users. Anthropic publicly disputed the severity of the government's threat finding, describing it as a 'narrow, non-universal jailbreak.' A bipartisan group of House members (Liccardo, Obernolte, Lieu, Franklin) sent a formal letter on 18 June 2026 demanding the legal basis for the directive, with a response deadline of 26 June 2026. By 22 June, legal analysts publicly characterised the directive as 'poorly drafted.' The Washington Post and CyberScoop confirmed that the Five Eyes joint statement context was directly linked to the Fable 5/Mythos 5 capabilities.
Why it matters
This is the first time the US government has used EAR authority to force a US AI lab to globally disable commercially released AI models on national-security grounds. It establishes a live precedent that EAR export-control tools can be applied to AI model access (not just hardware/chips). The directive's legal basis is contested, creating uncertainty for all US frontier AI developers about when the government may invoke similar authority against their models. The global compliance consequence — disabling all users, not just foreign nationals — demonstrates the practical extraterritorial reach of such directives.
Action needed
All US frontier AI developers should immediately review EAR applicability to their most capable models and establish compliance playbooks for rapid access-restriction in the event of a similar directive. Organisations relying on Fable 5/Mythos 5 should assess contingency model strategies. Monitor the 26 June 2026 Congressional deadline for Commerce Department legal-basis response, which may clarify or further constrain the directive's authority.