The US Department of Commerce dropped a regulatory hammer on Anthropic, ordering the company to block foreign nationals from accessing its two most advanced AI models. What followed was something nobody in Washington apparently planned for: Anthropic shut the models down for everyone.
On June 12, the Commerce Department issued an export control directive requiring Anthropic to cease access to Fable 5 and Mythos 5 for foreign nationals. The stated reason was national security, specifically a jailbreak vulnerability that officials determined posed risks to critical infrastructure. By June 13, Anthropic had disabled both models globally, concluding that selectively enforcing nationality-based restrictions was technically impractical.
From voluntary guidelines to forced shutdown in ten days
Just ten days before the Commerce Department’s directive, President Trump signed an executive order on AI innovation that emphasized voluntary cybersecurity collaboration for AI models. That June 2 order was supposed to set the tone for how the government would engage with frontier AI companies: gently, cooperatively, with a light regulatory touch.
The jailbreak vulnerability at the center of this decision reportedly allowed users to bypass safety guardrails in ways that could compromise critical infrastructure. Anthropic apparently determined that verifying the nationality of every user in real time, across every API call, wasn’t feasible. So instead of building a digital passport checkpoint, they pulled the plug entirely.
