Anthropic has taken down two of its AI models, Claude Fable 5 and Mythos 5, amid a U.S. government order citing national security concerns.

Following an export control directive issued on Friday afternoon, the company disabled access for all customers worldwide to ensure compliance with the order's vague parameters.

The move comes after earlier this year, when Anthropic faced restrictions from the Pentagon, which labeled it a 'supply chain risk.' The tech firm has since sued the Trump administration over these measures. Despite recent efforts to secure government trust through safeguards and limited rollouts, the current shutdown represents a significant setback for open AI innovation.

In a blog post, Anthropic argued that the jailbreak discovered by the U.S. government was narrow and not universal, but the company is still grappling with the implications of this directive on its operations and future projects.