The U.S. Commerce Department ordered Anthropic to suspend access to its Fable 5 and Mythos 5 AI models for all foreign nationals [1, 2, 3].
This directive marks a significant escalation in government oversight of artificial intelligence, signaling that the U.S. considers high-tier AI capabilities as critical national security assets that must be shielded from international actors.
The government order followed the discovery by researchers of a "jailbreak" — a method to bypass the safety guardrails designed to prevent the models from generating harmful content [1, 3]. This security vulnerability prompted the Commerce Department to intervene shortly after the technology became available to the public.
According to reports, the order arrived three days after the official launch of Fable 5 [4]. In a blog post, Anthropic said it received a government directive banning all foreign nationals from accessing Fable 5 and Mythos 5 [6].
There is some discrepancy regarding the full scope of the shutdown. While some reports indicate the ban specifically targets foreign nationals [5, 6], a company spokesperson said, "We will abruptly disable our most advanced AI models for all users" [1]. This suggests the company may have opted for a total suspension to ensure compliance with the security directive.
Other observers described the move as a forced action by Washington. One report said Washington forced Anthropic to disable its most powerful AI models over a disputed jailbreak [3]. The restriction applies to all foreign nationals worldwide [1].
“"We will abruptly disable our most advanced AI models for all users."”
This action establishes a precedent for the U.S. government to treat AI model access as a restricted export. By citing a 'jailbreak' as the catalyst, the Commerce Department is linking software vulnerabilities directly to national security risks, suggesting that the ability to bypass safety filters is now viewed as a strategic threat equivalent to the leak of sensitive military technology.
