Over the weekend, Anthropic took its two most advanced models — Fable 5 and Mythos 5 — offline at the White House's request. The company had little choice, according to statements, after the Trump administration demanded it block access for all foreign nationals, including employees abroad. The order was swift and imposed with minimal warning or public explanation.
Fable 5 and Mythos 5 were already subject to safeguards restricting their use in high-risk areas. The White House's decision to pull them entirely, according to reporting, stemmed partly from the belief that a group linked to China had accessed the Mythos model.
The shutdown has become a rallying point for governments worldwide arguing for technological independence from the US. In the UK, AI and Online Safety Minister Kanishka Narayan framed the incident as a sovereignty issue without directly naming the US or Trump. "We treat every other threat to our sovereignty with deadly seriousness, but we haven't learned to treat this one in the same way," he said, arguing that AI is "the central political question of our time" and that Britain must develop its own capacity before "someone else decides the answer for us."
France's response was more direct. Gabriel Attal, the Renaissance party's presidential candidate and former Prime Minister, called the shutdown the start of "the AI war" and drew a stark analogy: he likened Anthropic's model withdrawal to Iran's blockade of the Strait of Hormuz, characterizing access to frontier AI as a strategic chokepoint for which nations must prepare. Le Monde reported similar alarm across France's political spectrum, and members of the European Parliament have cited the Mythos and Fable shutdown as evidence that Europe must make tech sovereignty a priority.
Canada reached a similar conclusion. Prime Minister Mark Carney said the situation demonstrates the risk of relying on a single partner for access to critical resources. "The situation we're in collectively right now with Mythos and Fable is something that can happen with overreliance on certain models," he said. "Nobody has done anything wrong in the situation. But we will have done something wrong if we just accept this, don't take the lesson, don't build out and diversify."
The arguments for AI sovereignty are not new. Europe has spent years concerned about technological dependence on the US and has pushed for reduced reliance in chips, cloud computing, and AI. But the Anthropic shutdown has made the issue feel more urgent, compounded by broader unease about America's reliability as an ally under Trump, from trade disputes to threats regarding NATO.
China has long championed domestic AI firms and is among the few countries with models credibly rivaling American frontier labs, though some areas lag behind. Anthropic has accused Chinese rivals of using its models to train their own "on an industrial scale."
Most governments and businesses cannot match the resources of frontier labs in the US or China. However, sovereign AI does not necessarily require building the largest or most powerful tools. France's Mistral and Canada's Cohere demonstrate that solid efforts can emerge from outside these superpowers, even if they cannot compete directly. Other countries — Singapore and the UAE among them — have focused on narrower but still strategic priorities such as infrastructure or models optimized for local languages. Open-source models also represent a potential path forward.