Anthropic Halts Fable and Mythos AI Model Access After U.S. Government Directive
Anthropic has disabled access to its Fable 5 and Mythos 5 AI models following a directive from the U.S. Commerce Department. The order, issued under national security export controls, prohibits the company from distributing these models to any foreign national, including its own non-citizen employees both inside and outside the U.S. Anthropic described the government's position as a "misunderstanding" and stated it is working to restore access, confirming that its Claude models remain unaffected. Officials reportedly cited a technique capable of bypassing Fable 5's safeguards to access Mythos's cybersecurity capabilities. Anthropic countered that this was a narrow jailbreak, not universal, and believed similar capabilities could be elicited from other publicly available models not subject to such controls.
Anthropic was compelled to disable all access to its latest AI models, Fable 5 and Mythos 5, late on Friday. This action followed a directive from the U.S. Commerce Department, which utilized national security export controls to prohibit the company from distributing these models to any foreign national. The directive's scope encompasses individuals located outside the U.S. as well as foreign nationals within the U.S., including Anthropic’s non-citizen employees.
Given the broad nature of the order, Anthropic stated it had no alternative but to disable the models for all users. The company clarified that access to its less powerful Claude models, including the Claude Opus 4.8 model, remains unaffected. Anthropic communicated its apologies for the disruption to customers via a post on X, indicating its belief that the situation is a "misunderstanding" and that it is working to restore access promptly.
According to a blog post by Anthropic, the directive was received at 5:21 pm Eastern Time. The company noted that the letter it received did not initially provide specific details regarding the government’s national security concern. However, officials later informed Anthropic that the decision stemmed from the discovery of a technique designed to bypass Fable 5's safeguards. These safeguards were intended to prevent users from accessing the powerful cybersecurity abilities of Mythos, the foundational AI model for Fable 5.
Anthropic asserted its belief that the cited jailbreak was specific and would only unlock Mythos’s cybersecurity capabilities in one particular instance, rather than universally defeating all of Fable 5’s safeguards. The company also suggested that the same jailbreak could be used to elicit similar capabilities from other publicly available models, such as OpenAI’s GPT-5.5, which are not currently subject to comparable national security export controls.
Anthropic expressed disagreement with the premise that a narrow potential jailbreak should necessitate recalling a commercial model deployed to a large user base, stating that such a standard, if applied broadly, could halt all new model deployments for frontier model providers. The company also reiterated its stance that the government should possess the authority to block unsafe deployments, provided it is part of a statutory process that is transparent, fair, clear, and grounded in technical facts. Anthropic indicated that this particular action did not adhere to those principles.
Industry insiders and policy experts reportedly reacted to the directive with disbelief. Some observers linked the move to previous actions taken by the Trump administration against Anthropic. In February, U.S. President Donald Trump had reportedly ordered all federal agencies to cease using Anthropic’s models after the company declined to agree to the Pentagon’s preferred contract terms for AI vendors, which stipulated that purchased AI models could be used "for any lawful purpose." Anthropic had been seeking exemptions concerning the use of its models for autonomous weapons systems or mass domestic surveillance. The Pentagon subsequently declared Anthropic a "supply chain risk" in early March, instructing the U.S. military to stop using its models and prohibiting defense contractors from utilizing them for government contracts. Anthropic is challenging this designation in federal court.
Key Trump technology policy advisors, including former AI and crypto czar David Sacks and Pentagon undersecretary of defense for research and engineering Emil Michael, have publicly criticized Anthropic and its executives. Sacks has accused Anthropic of engaging in a "sophisticated regulatory capture strategy based on fear-mongering." Dean Ball, an AI policy expert with past service in the Trump administration who is critical of its recent decisions regarding Anthropic, commented on X that he could not determine if the action constituted "lawfare against Anthropic in particular or extreme national-security hawkery," describing it as "cartoonish."
(Source: Fortune)
