We believe this is a misunderstanding and are working to restore access
In a moment that reveals how swiftly the architecture of national security can reshape the landscape of artificial intelligence, the US government invoked export control authority to compel Anthropic to disable two of its AI models, Fable 5 and Mythos 5, for all foreign nationals worldwide. The directive, broad and immediate in its reach, forced a company built on global collaboration to choose between compliance and consequence. Anthropic, characterizing the order as a misunderstanding, now works to untangle the gap between government concern and operational reality — a negotiation that may define how AI companies navigate the deepening intersection of technology and sovereignty.
- A US government export control order landed without warning, demanding Anthropic shut down two flagship AI models on national security grounds — leaving no room for gradual compliance.
- The prohibition extended to every foreign national on earth, including employees physically sitting inside Anthropic's own offices, creating an overnight fracture in the company's workforce and customer base.
- Customers who had built operations around Fable 5 and Mythos 5 lost access entirely, as Anthropic was forced to disable the models globally rather than risk selective enforcement failures.
- Anthropic pushed back publicly, framing the directive as a misunderstanding — a carefully chosen word that signals the company believes the government acted on flawed or incomplete information.
- Negotiations to restore access are now underway, but the timeline, the evidence required, and the government's willingness to reconsider all remain uncertain, leaving users and employees in suspension.
The US government ordered Anthropic to shut down two of its AI models, Fable 5 and Mythos 5, citing national security concerns under export control authority. The directive was sweeping: no foreign national, regardless of where they lived or worked, could access the systems. For Anthropic's leadership, the choice was stark — comply immediately or face legal consequences.
The company announced the shutdown publicly, explaining that compliance required disabling both models for its entire global customer base. There was no surgical option. International clients lost access to tools they depended on, and foreign national employees — including those working inside Anthropic's US offices — found themselves locked out of company systems. Anthropic apologized for the disruption, but the damage was already done.
What distinguished Anthropic's response was its framing of the order as a misunderstanding — a word that implies the government may have acted on incomplete or inaccurate information about the models or their applications. The company said it is actively working to restore access, suggesting that clarifications or negotiations with US authorities are already in motion.
The episode stands as a sharp illustration of a broader shift: export control authority, long reserved for weapons and sensitive dual-use technology, is now being applied directly to artificial intelligence — with consequences that are immediate, global, and deeply human.
The US government has ordered Anthropic to shut down two of its artificial intelligence models, Fable 5 and Mythos 5, citing national security concerns. The directive, issued under export control authority, bars any foreign national from accessing the systems—whether they are located abroad or working inside the United States. For Anthropic, led by Dario Amodei, this meant an abrupt choice: comply immediately or face legal consequences.
The company announced the shutdown on X, explaining that the order left them no practical alternative. To ensure compliance with the export control directive, they had to disable both models entirely for their entire customer base. This was not a selective restriction targeting certain countries or regions. It was a blanket prohibition that affected everyone—international clients, foreign employees, and any user outside US borders.
The disruption was immediate and sweeping. Customers who relied on Fable 5 and Mythos 5 for their operations suddenly lost access. Foreign national employees at Anthropic, including those physically present in the company's offices, could no longer use the systems. The company acknowledged the damage in its statement, apologizing for what it called a disruption to its customers.
But Anthropic's response carried a notable qualifier. The company characterized the order as a misunderstanding—a suggestion that the government may have acted on incomplete or inaccurate information about the models, their capabilities, or their use cases. This framing hints at a path forward: if the misunderstanding can be clarified, the order might be reversed or modified.
The company said it is working to restore access as soon as possible, signaling that negotiations or clarifications with US authorities are underway. What remains unclear is how long that process will take, what evidence or arguments Anthropic will present to challenge the order, or whether the government will reconsider its position. The shutdown stands as a stark illustration of how export control authority—traditionally used to restrict military or sensitive dual-use technology—is now being applied to artificial intelligence systems, with immediate and tangible consequences for a major AI company's operations and workforce.
Notable Quotes
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.— Anthropic statement
We apologise for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.— Anthropic
The Hearth Conversation Another angle on the story
Why would the US government issue an export control order on AI models? What's the national security angle here?
Export controls are typically used to prevent sensitive technology from reaching adversaries or being used in ways that threaten national security. With AI, the concern is that advanced models could be used for purposes the government wants to restrict—whether that's military applications, surveillance, or other sensitive uses. The government is essentially treating frontier AI like it treats advanced semiconductors or weapons technology.
But Anthropic is a US company. Why block foreign nationals from accessing models made by an American firm?
That's the blunt instrument part of it. Export controls don't just restrict what leaves the country—they restrict who can access it, period. A foreign national employee at Anthropic's US office is still a foreign national in the eyes of export law. The government is trying to prevent any foreign person from touching the technology, regardless of where they are.
Did Anthropic do something wrong, or is this really a misunderstanding as they claim?
The company's language suggests they believe the order was issued based on incomplete information—maybe the government didn't fully understand what these models do, or thought they had capabilities they don't actually have. Anthropic is essentially saying: if you understood what we actually built, you wouldn't have done this. That's a negotiating position, not a confession.
What happens to the customers who were using these models?
They're cut off. No access, no warning, no gradual transition. That's the cost of export compliance. Anthropic has to choose between breaking the law and breaking its customer relationships. They chose the law.
Is this the future of AI regulation?
It might be. Governments are increasingly treating AI as a strategic technology. This order shows that export controls—a Cold War-era tool—are now being weaponized for AI. If other countries follow suit, we could see a fragmented global AI landscape where different models are available in different regions, or where international collaboration becomes much harder.