Anthropic Halts Access to Fable 5 and Mythos 5 After US Order

Anthropic has suspended access to its newly released Fable 5 and Mythos 5 AI models after receiving a U.S. government export control directive, citing national security concerns. The company disabled the models for all users immediately to comply with the order, while saying its other offerings—including Opus 4.8—remain available.

In a statement posted Friday, Anthropic said the directive arrived at 5:21 pm ET and instructed it to suspend “all access” to Fable 5 and Mythos 5 for any foreign national, whether inside or outside the United States. This restriction reportedly includes foreign national Anthropic employees, and the company said it took broad action to ensure compliance.

Key takeaways

• Anthropic suspended access to Fable 5 and Mythos 5 immediately after receiving a U.S. government export control directive.
• The order reportedly targets access by foreign nationals, including Anthropic employees who are foreign nationals.
• Other Anthropic models, such as Opus 4.8, are not affected according to the company.
• Anthropic said authorities raised concerns about a potential “jailbreak” method that could bypass safeguards on Fable 5.
• The firm described the government’s evidence as “verbal” and suggested the issue involves a narrow, non-universal jailbreak rather than a broad one.

Export control directive triggers immediate model shutdown

Anthropic’s action follows an abrupt interruption to access for the public. According to the company, it received the directive late Friday and was told to suspend access to Fable 5 and Mythos 5 by any foreign national. To meet the requirement without exception, Anthropic said it removed access for all users rather than attempting to segment access by nationality.

The company framed the move as a straightforward compliance step: it is “removing access to Fable 5 and Mythos 5 for all users” to comply with the government’s legal directive.

Why Anthropic says the concern is limited

While Anthropic did not provide specific details about the alleged threat, it said it believes the government is concerned about a possible jailbreak technique capable of bypassing safeguards built into Fable 5.

In its statement, Anthropic noted that, to date, the government has provided only verbal evidence of a potential “narrow, non-universal jailbreak.” The company described this as essentially asking the model to read a specific codebase and fix software flaws—an approach it argued is materially different from a “universal jailbreak,” which would broadly undermine protections across scenarios.

Anthropic also pushed back on the severity of the response implied by the order. The firm said it “disagree[s]” that a narrow potential jailbreak should lead to the recall of a commercial model deployed at large scale. It added that applying that standard across the industry would effectively stop new frontier model deployments for all providers.

Recent release raises questions for AI users and operators

Anthropic’s suspension comes only days after it released both Fable 5 and Mythos 5. The releases were notable not just for their capabilities, but for the underlying context around Mythos Preview, which Anthropic previously said had helped uncover thousands of vulnerabilities in critical software.

Earlier coverage around these releases highlighted the scale and intensity of the safety research and testing that can surround frontier model rollouts—particularly when models are capable of complex reasoning and code-related tasks. In that setting, the sudden reversal underscores how quickly external compliance actions can override product continuity.

Anthropic also indicated that it believes the government order is the result of a misunderstanding and that it is working to restore access for users “as soon as possible.” For model users—especially those outside the U.S.—the key near-term issue is whether access can return in a way that matches the directive’s scope without requiring a full shutdown.

What to watch next

Until Anthropic receives clearer guidance or the government narrows the directive’s implementation, users should expect continuing uncertainty around when Fable 5 and Mythos 5 will be available again and under what geographic or eligibility conditions. Investors and builders in the AI sector will likely watch closely for how regulators distinguish between narrow jailbreak techniques and broader safeguard failures—and whether the incident prompts tighter deployment controls across the industry.