An export control directive announced by the U.S. has forced Anthropic to suspend access to its Claude Fable 5 and Mythos 5 models for all foreign nationals, prompting sharp criticism from across the tech world that the policy is unworkable, risks harming Amer
A fight over who gets to use the newest frontier models broke into the open after Anthropic said it had received an “export control directive” that shut off access to its Claude Fable 5 and Mythos 5 for foreign nationals—an order the company said has left even its own non-U.S. employees locked out of tools they built.
The directive. announced Friday by the California-based company. prevents use “by any foreign national. whether inside or outside the United States. including foreign national Anthropic employees.” Anthropic said the government provided limited details. but the company believed the action followed a potential vulnerability—specifically a method of “jailbreaking” Fable 5.
Jailbreaking refers to bypassing software restrictions such as safety guardrails designed to protect a system or network. Anthropic said it investigated the issue and concluded the exposed problems were limited to a handful of previously known weaknesses that other publicly available models can also discover. The company also acknowledged that “watertight jailbreak protection is likely ‘not currently possible for any model provider.’”.
Because of the order, Anthropic said it had to suspend all access to both models to comply. The company apologized to its customers for the “disruption. ” and said it believed the directive stemmed from a “misunderstanding” by the government. Newsweek has contacted Anthropic and the Commerce Department via email for comment.
Critics across the tech industry pushed back hard on the premise that a narrow. security-related concern should lead to an across-the-board ban on foreign use. In a statement posted Friday. Anthropic said. “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.” It added that if the same standard were applied across the industry. it would “essentially halt all new model deployments for all frontier model providers.”.
Dean Ball. a former White House AI adviser and one of the authors of the Trump administrations “AI Action Plan. ” described the decision as “baffling.” In posts on X over the weekend. Ball questioned how the U.S. could seek to restrict foreign access to Anthropic models while also maintaining a posture of exporting advanced AI chips to China. He wrote on Saturday: “An administration whose posture is that we *should* export advanced AI chips to China. which also wants to ban… Britain (and every other non-American on Earth)… from using our best models?”.
Ball added in another post that he could not tell whether the move was “lawfare against Anthropic in particular or extreme national-security hawkery,” calling it “simply cartoonish.”
Chris McGuire. senior fellow for China and emerging technologies at the Council on Foreign Relations and a former State Department adviser. said targeted export controls on AI models could be “prudent. ” but criticized what he described as an attempted universal ban. He said “Imposing equally broad deemed export controls. which also restrict access to foreign nationals. is just absurd—and obviously will result in the model being pulled from distribution. as just happened.” McGuire. who also served as a National Security Council adviser during Joe Biden’s tenure. questioned how the U.S. could export AI chips to China while restricting access to “the models that those same chips produce and run.”.
Other critics argued the ban landed in a place Anthropic helped shape. Peter Girnus. a software threat researcher. said part of the responsibility lies with the company because it marketed its Claude Mythos model as too powerful to release. “If you describe your product as a munition in every press release. eventually a government takes you at your word. ” Girnus wrote on X. He said the legal framing was effectively self-made: “They wrote the legal predicate themselves and called it a brand.”.
Still, Girnus argued export controls like these would struggle because of the nature of what’s being restricted. “Math does not stop at customs,” he wrote, adding that the order left Anthropic’s own foreign-national employees “locked out of the model they built.”
Gary Marcus. a scientist known for his work and frequent commentary on AI. called the directive “shocking” and said it would be “counterproductive for the US AI industry.” He posted on X that it could help China instead. writing: “Perhaps it does China a favor. though.” He also said “every Chinese person working in a US AI company (and there are many) will consider returning to the competition in China ASAP. ” and argued investors would begin to question whether American AI firms can thrive “in this atmosphere.”.
The backlash extended beyond the U.S. Vassant Shetty. an Indian entrepreneur. said a universal export control covering allies and adversaries alike would create friction between global powers. In a post on X. he pointed to India’s role as “the second-largest market globally for both ChatGPT and Anthropic. ” and warned that if access can be turned off “at the press of a button like this. ” users would be “absolutely at the mercy of a foreign govt.” He wrote that “Globalisation in the current form is dead. ” calling it “a huge wakeup call for India.”.
Ciaran Martin. a former chief of the UK government’s National Cyber Security Centre. said the development “came completely out of the blue” and predicted it would “probably” prove “unsustainable.” Speaking during an interview with Channel 4. Martin argued that while the order is framed as an attempt to stop foreign nationals accessing a powerful model on security grounds. the only practical way to do it is to withdraw the model completely. He said Anthropic employs “thousands of foreign nationals,” and serves “millions of users” whose nationality can’t be detected. Martin added that the U.S. appears “obsessed with winning the AI race with China,” and that this “slows it down.”.
The sequence of concerns from multiple angles—security logic. industrial consequences. and the sheer difficulty of enforcement—has converged on the same point: a control meant to address a vulnerability has translated into an access ban so wide it reaches the people inside the company who are themselves foreign nationals.
Anthropic has faced repeated tensions with the U.S. government, including disputes involving the military applications of its technology and its potential use for mass domestic surveillance. Beyond this export-control effort aimed at Fable 5, the company is also in a legal battle with the Department of Defense. Anthropic was placed on a supply chain blacklist earlier this year after refusing to allow the military unrestricted access to its Claude models.
Many in the tech space believe the government will back down or amend the directive to block foreign nationals from using the models. Anthropic said it was appealing the order. In its Friday statement. the company said it believes 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.” It said the action “does not adhere to those principles.”.
“We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible,” Anthropic said.
Trump export control directive Anthropic Claude Fable 5 Mythos 5 jailbreaking AI regulation China Department of Commerce Department of Defense national security