The U.S. prohibits non-Americans from using Fable 5 and Mythos 5 models.

It was reported that the United States Department of Commerce had ordered Anthropic to prohibit any foreign citizen from visiting its latest and most powerful AI model. Anthropic stated that, in order to comply with the directive, access to Fable 5 and Mythos 5 had to be restricted for all users (not just foreigners).

In a statement issued on Friday evening, Anthropic wrote: “The United States Government has issued an export control order to suspend access to Fable 5 and Mythos 5 of all foreign citizens, whether or not they are in the United States.” Anthropic’s decision to suspend user access appears to be the first time that the AI giant would be publicly deployed under a model that would fall offline as a result of federal government intervention.

“The end result of this order is that we must immediately ban all client Fable 5 and Mythos 5 services to ensure compliance”, Anthropic writes. Currently, in the Claude Network or API call, the existing dialogue on the original Fable 5 has been terminated by error and the new session has been forced to return automatically to the default model or Opus 4.8. Basic models such as Sonnet and Haiku are not affected by this ban.

According to Anthropic, the United States Government has discovered a specific technology that may allow users to bypass part of Fable 5 ‘ s safety precautions. Anthropic also claims that the Government has provided only “verbal evidence” on the issue and that this circumvention may also exist in the AI system developed by other companies.

Anthropic emphasizes that at this stage no manufacturer is able to escape from prison with 100 per cent immunization, which is not universal. If the U.S. Government were to force the fall of the product on the sole basis of a very limited, potential escape demonstration, it would effectively paralyse the landing of new models by all forward model manufacturers across the industry.

On Tuesday, Anthropic released the company’s most powerful AI models so far – Fable 5 and Mythos 5. Anthropic states that these AI models are powerful and therefore require strict safety precautions to prevent malicious or dangerous use. “It is risky to publish such a powerful model. In the absence of security measures, Fable 5 may be misused in areas such as cybersecurity, causing serious damage”.

Although Fable 5 and Mythos 5 are based on the same technology base, only Fable 5 is open to the public. The Fable model places stricter limits on the types of questions that users can ask, especially on issues related to cybersecurity and biology. In the absence of such protective measures, Mythos 5 was issued only to some trusted partners such as Cybersecurity and Infrastructure.

Anthropic officially stated that this was a serious “technical misunderstanding” and that it was conducting urgent high-level good offices with the White House and the Ministry of Commerce, promising to publish more details within the next 24 hours and making every effort to resume the visit at an early date.