US government orders Anthropic to suspend Fable 5 and Mythos 5 access over export controls

In a June 13 statement, Anthropic said the US government 'citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.' Earlier reporting from Wired and TechCrunch said the government believed it had found a method of bypassing or 'jailbreaking' Fable 5, prompting the order to halt access to the top-end models. A Medium analysis framed the action as tied explicitly to export-control measures governing frontier-model deployment.

The mechanism is striking: rather than a recall over a model defect, this is a state security action against a commercially deployed AI, invoking the same export-control regime historically used for advanced chips and dual-use technology. The directive's foreign-national scope — reaching even Anthropic's own non-US staff — suggests the government is treating model weights and access as controlled technology.

The irony is sharp and was noted across the community: Anthropic has been the industry's loudest safety advocate, and CEO Dario Amodei argued on June 10 that frontier models should face 'mandatory third-party testing for cyber, bio, and autonomy risks — with the power to block or revoke deployment.' Days later, the government did exactly that to Anthropic's own models. TechCrunch's headline captured it: 'Anthropic's safety warnings may have just backfired.'

Watch for whether the suspension expands to API customers broadly or stays scoped to foreign nationals, and how Bedrock and GitHub Copilot — which had only just gained Fable 5 — handle the disruption. The precedent matters: if Washington can revoke a deployed model over a jailbreak claim, every frontier lab now faces regulatory deployment risk it hadn't priced in.