US order forces Anthropic to pull 2 AI models

On June 12, 2026, the U.S. government issued an export-control directive focused on Anthropic’s newest artificial intelligence models, Fable 5 and Mythos 5. The order requires suspension of access to these models for any foreign national, whether they are inside or outside the United States. Foreign nationals employed by Anthropic are also covered by the directive. Officials cited national security authorities as the legal basis for the action.

The directive arrived at Anthropic at 5:21 p.m. ET on June 12, according to the company. It specified that access to Fable 5 and Mythos 5 must be cut off for foreign nationals, but did not impose similar restrictions on Anthropic’s other models. The order is framed explicitly as an export-control measure rather than a broader ban on the company’s technology.

To ensure immediate and complete compliance, Anthropic responded by abruptly disabling access to Fable 5 and Mythos 5 for all customers, not only those identified as foreign nationals. The company announced that both models were taken offline shortly after receipt of the directive. This action affects every user of the two models, including domestic customers in the United States.

Anthropic stated that its remaining product lineup is not impacted by the government’s order. Access to all other Anthropic models continues to operate normally for customers. By limiting changes to Fable 5 and Mythos 5, the company has tried to confine disruption to its newest offerings while keeping existing services available.

Anthropic said the government’s concerns stem from a technique officials believe can bypass or “jailbreak” some safeguards in Fable 5. The company reviewed a demonstration of a specific method and acknowledged that it can expose certain vulnerabilities in the model. However, Anthropic characterized these issues as a small number of minor, previously known vulnerabilities.

The company also said that similar capabilities exist in other publicly available AI models. It cited other leading systems as examples of models where comparable vulnerabilities have been observed. Anthropic noted that the letter conveying the directive did not include detailed explanations of the underlying national security concern, beyond the reference to a potential jailbreak technique.

Reporting on the directive indicates that the letter was issued by Commerce Secretary Howard Lutnick. The Commerce Department’s Bureau of Industry and Security was described as having helped prepare the order, reflecting the use of established export-control mechanisms. This places the action within the broader framework of U.S. oversight of sensitive technologies.

Anthropic’s removal of Fable 5 and Mythos 5 from public access appears to mark the first known instance of a leading AI company taking deployed models offline as a direct result of federal government intervention. The move highlights how export-control tools are being applied to advanced AI systems and underscores the compliance steps providers may take when faced with sudden restrictions on foreign access.