SK By Stuart Kerr, Technology Correspondent, LiveAIWire
On June 12, 2026, a single letter from the US Commerce Department took two of the world’s most capable AI models offline for every user on the planet within hours. On July 1, the same department gave them back. Nineteen days separated the shutdown from the restoration, and in that window, governments from Paris to Ottawa to Vienna reached the same conclusion: the country that hosts the world’s leading AI labs can unplug them from the rest of the world on a private letter, with no hearing, no warning and no vote. The reversal did not change that fact. It confirmed it.
The models in question were Claude Fable 5 and Claude Mythos 5, Anthropic’s most advanced systems, released to the public on June 9. Three days later, the Commerce Department’s Bureau of Industry and Security invoked export control law, ordinarily reserved for physical hardware like semiconductors, to order Anthropic to cut off access for any foreign national anywhere, including the company’s own non-citizen staff. Because Anthropic had no way to verify a user’s nationality in real time, it disabled both models for everyone worldwide rather than risk violating the order. The stated trigger, according to Anthropic’s own statement on the directive, was a disputed jailbreaking technique that the company said produced only minor, already-public vulnerability findings, comparable to what other publicly available models could already do.
Why Allies Reacted Like They Had Been Cut Off, Not Just Denied a Product
The scale of the diplomatic reaction is the part of this story that outlasted the shutdown itself. As Al Jazeera reported from the G7 summit, French President Emmanuel Macron told the gathering days after the order that the move was a wake up call about AI dependence, calling the reaction “strictly nationalist.” Canadian Prime Minister Mark Carney warned that accepting single country dependence on US AI infrastructure would itself become the mistake if allies failed to diversify. Former UK security minister Tom Tugendhat wrote that the episode showed sovereignty was becoming a matter of code rather than cannons. None of these were routine trade complaints. They were G7 leaders describing a commercial product shutdown in the language normally reserved for energy or defence dependence.
That reaction did not fade when access came back. Austria has since formally asked the European Commission to explore hosting Anthropic within EU borders, arguing that only relocating the legal jurisdiction that controls a model’s access, not the company’s headquarters, would put the killswitch out of Washington’s reach. The proposal is unlikely to happen given how deeply Anthropic’s compute, capital and cloud partnerships are anchored in the United States. But the fact that a national government formally raised it inside three weeks of the shutdown tells you how seriously European capitals are now pricing in the risk.
The Precedent Did Not Start Here, and That Is What Makes It Believable
Governments outside the US were primed to read the Anthropic order as a pattern rather than an anomaly because a version of this had already happened with hardware. Washington spent much of the past three years tightening and loosening chip export rules aimed at China, including the on-again, off-again restrictions around Nvidia’s advanced AI chips, which showed allies and rivals alike that US export authority over compute could be switched on or off depending on the political moment. What changed on June 12 was the target. For the first time, that authority reached beyond physical chips to control access to a deployed software model, opening a legal question that specialists have since described as plausible but genuinely untested: whether serving an AI model from a US company’s own servers to a customer abroad even counts as an export in the first place.
The timing compounded the anxiety. The order landed one week before the heads of the Five Eyes cybersecurity agencies issued a joint warning that frontier AI would reshape the cyberattack landscape within months, not years. Read together, the two events told the same story from opposite directions: intelligence agencies were telling the world that powerful AI was becoming critical security infrastructure, just as one of those agencies’ own governments demonstrated it could remove a piece of that infrastructure from the rest of the world overnight.
What This Means for You
If your business runs workflows, products or research on a frontier US model, the practical lesson of June and July is not that the models are unsafe. It is that access itself is now a variable outside your control, subject to a government decision you will not see coming and cannot appeal. Anthropic has since proposed an industry wide framework, alongside Amazon, Microsoft and Google, for scoring jailbreak severity so that future disputes are resolved against agreed criteria rather than a single unexplained letter, an acknowledgment from inside the industry that the current process left everyone exposed. Until frameworks like that are tested, any organisation building on frontier AI outside the US should treat model access the way it treats a single supplier relationship in a critical supply chain: worth mapping, worth having a contingency for, and no longer worth assuming is permanent.
The Chilling Effect Is Already Showing Up in Where Money and Attention Are Going
The clearest sign that the precedent has outlived the three week shutdown is where policy attention has moved since. Germany and France’s domestic intelligence services have shifted contracts away from some US analytics providers toward European alternatives, a decoupling that began before the Anthropic order but accelerated after it. The European Commission’s spokesperson for tech sovereignty told reporters that the response to security concerns should be a shared challenge rather than a decision made by a single jurisdiction, language that signals Brussels sees its own forthcoming AI and cybersecurity rules as insurance against a repeat. None of this reflects a rejection of American AI. It reflects a judgment, now made openly by sitting heads of government, that treating any single country’s models as an unquestioned default was itself the risk.
Why the Restoration Does Not Reset the Clock
Commerce Secretary Howard Lutnick framed the July 1 reversal as the product of close coordination with Anthropic to strengthen safeguards, and the company has indeed added new protections before Fable 5’s return. That is a real resolution, and it matters that the models are back. But resolution of one dispute does not retract the authority that produced it. The legal theory that let Commerce reach a software model rather than a physical export has not been tested in court, has not been subjected to formal rulemaking, and remains available for the next disputed finding, whenever and against whichever company that turns out to be. Every government and company that watched June 12 happen now knows the tool exists. That knowledge, not the three weeks the models spent offline, is the story that will keep shaping AI policy decisions in capitals that have no vote in Washington.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, emerging technology, and their impact on business, society, and geopolitics. LiveAIWire publishes daily AI news and analysis at liveaiwire.com.