US Government Bars Foreign Access to Anthropic's Fable 5 and Mythos 5

The US government has issued an unprecedented export control order blocking all foreign nationals—even those working inside Anthropic—from accessing its two most advanced publicly available AI models: Fable 5 and Mythos 5. The order, citing national security concerns, targets a potential 'jailbreak' technique where an AI is asked to read and fix software code vulnerabilities. Anthropic pushes back, arguing this capability is not unique; other models like OpenAI's GPT-5.5 perform similar tasks, and the feature is routinely used by cybersecurity professionals. This is the first time export control law has been applied directly to an AI software product rather than physical goods.

This is not Anthropic's first friction with US authorities. In March 2026, the Pentagon labeled Anthropic a 'supply chain risk' after the company insisted its AI (Claude) should not be used for mass domestic surveillance or fully autonomous weapons. The company is reportedly considering legal challenges to that label. The export control order arrives as Anthropic prepares for a public stock listing, with a recent valuation of $965 billion. Both Fable 5 and Mythos 5 are built on the new 'Mythos-class' architecture, derived from the earlier Mythos Preview model—which was not publicly released because it could identify severe vulnerabilities in major operating systems and browsers. The released models include additional safety features.