U.S. export – Anthropic says it has been ordered to suspend access to its Fable 5 and Mythos 5 AI models for all foreign nationals, including outside the United States and even inside the country. The move arrives days after the company filed for an IPO, leaving investors a
For the second weekend in a row, Claude users hoping to keep testing Anthropic’s newest models are hitting a wall—one that wasn’t there when they went to bed Friday night.
The U.S. government issued an export control directive that suspends all access to Anthropic’s Fable 5 and Mythos 5 models by any foreign national. whether they are inside or outside the United States. Anthropic says the order also reaches the company’s own employees. To comply, the company says it will have to suspend access to the models for all customers.
The timing is especially painful for a company that has just begun pitching itself to the public market. Anthropic recently filed paperwork for a proposed initial public offering of its common stock. Investors have been watching other major AI companies—SpaceX and OpenAI among them—talk openly about their own IPO ambitions. Now, Anthropic’s models are effectively frozen for a broad swath of users.
The directive also lands in the middle of a hard question: why block Mythos, and why now?
In a blog post, Anthropic said the government’s order did not specify the national security concern. But the company said it believes the Trump administration became aware of a way people can bypass security protocols on the Fable 5 model. Anthropic described the vulnerabilities as “previously known” and “minor.”.
The company wrote that the vulnerabilities appear “relatively simple. ” and that it had found other publicly available models can discover them without requiring a bypass. Anthropic also said it was deliberate about how it released Fable and Mythos. including providing major corporations and the government with early access to the model to test its usage.
Anthropic said it built “strong safeguards” to reduce the likelihood that people could misuse Fable for tasks related to cybersecurity. The company added that many users have complained those safeguards are “overly broad.”
Company officials also said no one has been able to find a universal method to “jailbreak,” or break through those safeguards, which would otherwise unlock a large array of cyber capabilities.
The export control mechanics were sharpened on Friday in a letter Commerce Secretary Howard Lutnick sent to Anthropic CEO Dario Amodei. The letter. as described by Axios. says the models would be subject to export controls for any location outside the U.S. and for all foreigners within the country. The same outlet reported that the administration tried to get Anthropic to pause the models’ release but was unsuccessful.
An official told Axios that the models would remain locked until the U.S. government’s national security apparatus is stronger. The official said it could take a few weeks.
Anthropic is not arguing that governments should ignore risks. The company says it previously agreed that the government should be able to block unsafe deployments. But officials say they disagree with the way the Trump administration is handling the decision. arguing that if the same reasoning were applied across the AI industry. “it would essentially halt all new model deployments for all frontier model providers.”.
“When people are told their access is being cut off—especially during a high-profile moment like an IPO filing—it doesn’t just affect a product schedule. ” one investor-facing reality hangs in the background: regulatory uncertainty becomes a business risk. Anthropic’s export control order could complicate the company’s listing plans because investors may see it as a sign of unpredictability around whether frontier models can be deployed quickly.
The company’s dispute with the administration also comes with a political undertone. Anthropic’s critics inside the Trump orbit have described the company as “woke” and “leftist. ” and the administration’s decision has fed questions about whether politics is mixing into what is supposed to be an export-control evaluation grounded in technical concerns.
This is not the first time Anthropic has clashed with U.S. power. The company’s earlier ethics—and its refusal to give the Pentagon full control over its AI models—triggered a backlash. Anthropic cited its own rules on AI weapons use. The U.S. military responded by blacklisting Anthropic and prohibiting any companies connected with the U.S. government from accessing it.
Now, Anthropic is back in the line of fire—not over deployment by U.S. partners, but over access to its newest frontier models.
Anthropic says the latest action is not what it signed up for when it supported the idea of transparent safety steps. In the company’s view. the government should be able to block unsafe deployments as part of a statutory process that is transparent. fair. clear. and grounded in technical facts. The company said it believes the current step “does not adhere to those principles.”.
Anthropic Fable 5 Mythos 5 export controls Howard Lutnick Dario Amodei IPO Claude United States politics national security AI regulation cybersecurity safeguards