Anthropic announced it will cut off access to its newest leading AI systems, known as Fable 5 and Mythos 5, in response to a US government directive restricting foreign access to the technology.

The firm received official notification late Friday afternoon around 5:21 p.m. Eastern Time, highlighting potential threats to national security tied to its AI offerings.

According to the order, the restrictions apply to any foreign nationals both within and outside the United States, encompassing even Anthropic's own overseas staff. As a result, the company plans to deactivate the models universally to maintain full compliance.

The communication from authorities omitted particular explanations for the security worries.

Anthropic suspects the primary issue involves a specific method to circumvent safeguards on Fable 5, though it argues the problem is limited in scope and relates to already documented weaknesses that other open models could also expose.

This action represents a further intensification in ongoing tensions between Anthropic and the Trump administration concerning AI regulations, safety protocols, and governmental oversight of cutting-edge models.

Back in February, the Pentagon labeled Anthropic a potential supply chain vulnerability after the company attempted to impose restrictions on its AI for certain military uses.

The startup has taken legal action against the Defense Department regarding this classification, with two related cases still unresolved in court.

A White House representative offered no immediate statement. Meanwhile, the Pentagon's top information officer voiced approval via social media, noting that national security takes precedence over business interests.

Anthropic confirmed it is following the directive despite disagreeing with the assessment. The firm has not specified the exact timing for the shutdown. Its remaining AI models will continue operating normally.