Just three days after their launch, Anthropic has disabled global access to its two most advanced AI models — Claude Fable 5 and Claude Mythos 5 — after the U.S. government issued an export control directive citing national security concerns.
According to Reuters, Anthropic said it had been ordered to suspend access for all foreign nationals. In practice, the shutdown went further: according to Engadget and Anthropic's own status page, all customer access worldwide was suspended.
The trigger was a jailbreak. According to Wired, Anthropic's blog post stated that "the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5." According to Ars Technica, the Commerce Department specifically worried that such a jailbreak could become a national security threat, with a separate report from CyberSecurityNews noting the model had been jailbroken to generate stack exploits. According to Axios, Commerce Secretary Howard Lutnick personally informed Anthropic that the models would be subject to export controls after a company claimed it could jailbreak Mythos.
Anthropic pushed back publicly. According to Reuters, the company said it believes the order is a "misunderstanding" and is working to restore access. In its blog post, the company wrote that it disagrees "that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," according to TechCrunch. According to Financial Express, Anthropic also noted that authorities "did not provide specific details of its national security concern."
AWS confirmed that all other Anthropic models remain unaffected.
This appears to be the first time U.S. authorities have used export control powers to force an AI company to take a flagship commercial model offline — a move that could fundamentally change how the industry handles security disclosures and government oversight of AI.