Anthropic export – Anthropic says a Trump administration order forced it to block access to its Fable 5 and Mythos 5 models for all foreign nationals, including people inside the US and Anthropic employees. The company cites “national security authorities” behind an “export cont
For much of the week, Anthropic has been fighting to get its newest AI models back online. The problem wasn’t a technical failure or a capacity crunch. It was access—specifically, who was allowed to use its models.
The Trump administration abruptly ordered Anthropic to cut access for all foreign nationals, including users inside the US and the company’s own employees. Anthropic responded by blocking access to its Fable 5 and Mythos 5 models for everyone.
In a statement on its website. Anthropic said the government cited “national security authorities” to justify an “export control directive” on the models. The company also said the administration’s concerns about a “jailbreak” potentially used by groups linked to China to access its models did not allow users to circumvent all of the company’s safeguards.
The public explanation, at least from the administration, hasn’t arrived. That absence has left even experts wrestling with a basic question: what is the US actually trying to “export” here?
Hanna Dohmen. a senior research analyst at Georgetown University’s Center for Security and Emerging Technology. described the move as unprecedented. “To my knowledge. this is the first time US export controls have been used to control access to an AI model in this way.” She called it “an open question” whether the order strains existing rules without seeing the precise language behind it.
Export controls are usually built for things that can be shipped across borders—weapons, hardware, tools, and related categories. Over time. the framework expanded to cover less tangible items such as software. source code. technical data. and even 3D-printed gun files. In those cases, the regulated material is something that can be copied, downloaded, published, or otherwise handed over.
Anthropic’s situation doesn’t fit neatly into that model. Mythos and Fable are still hosted on Anthropic’s servers. Users aren’t receiving source code, model weights, or a copy of the model itself. They’re getting chatbot responses to their queries.
The export could be some specific information produced by the models—but it’s not clear. in that scenario. why the government would disable access to an entire system rather than limiting a part of it. It could also be access itself. Yet remote access to cloud services has been a known gap in export control regimes. and Congress is already trying to close it through legislation now moving through the Senate.
The confusion extends beyond legal mechanics and into what the industry is supposed to do next. Andrew Reddie. a professor at UC Berkeley’s Goldman School of Public Policy. said export controls and other regimes like arms regulations give the government “wide latitude” to restrict access to certain goods. But he argued that “the equivocation by successive administrations regarding what the responsibilities of model developers are” has made it hard for firms to understand what is expected.
Reddie said the episode is notable because it points to wide uncertainty about the responsibilities of companies building frontier AI systems.
If the order targeted Anthropic because Mythos and Fable are uniquely capable. experts say it raises a straightforward question for the next wave of models from OpenAI. Google. Meta. xAI. and other frontier labs: what standard will trigger access cuts next time?. If the models were singled out due to safeguard issues. then the administration still needs to say what level of protection it considers sufficient. And if Anthropic was targeted because its relationship with the Trump administration is politically tense. the rationale becomes even harder to square with a consistent regulatory approach.
“This episode makes clear the unsustainability of the existing governance regime.”
Experts also warned about what such unpredictability means for global trust. They said the incident has already added fuel to concerns outside the US about relying on American companies for access to strategically important systems.
Reddie’s concern was tied to a specific pressure point: whether the government is primarily worried about jailbreaks—users bypassing safeguards. He said, “If creating models that are impossible to jailbreak becomes the de facto standard for the United States, then it will have no AI models.”
Underneath the debate about legality and safeguards sits a harder political reality. The Trump administration has repeatedly signaled a hands-off approach while championing American technology. Yet Anthropic was forced to yank its frontier models through an order it said has not been publicly explained.
The upshot is a mismatch between aspiration and implementation. If Washington wants to control who can access powerful AI systems. experts argue it needs to clarify how that control works and allow firms a real chance to comply before models launch. Ad hoc interventions delivered without clear standards are not sustainable—and. experts say. they risk putting the United States behind in the race to set the terms of frontier AI.
For now. the immediate human effect is plain: Anthropic blocked access to Fable 5 and Mythos 5 for everyone after the order required cuts for foreign nationals. including people in the US and its own employees. And the bigger uncertainty remains unresolved—what the export control order actually covers. and how companies are expected to navigate it before the next model release becomes another test.
Anthropic Fable 5 Mythos 5 Trump administration export controls AI governance national security authorities jailbreaks cybersecurity policy frontier AI