A US export ban on Anthropic’s most advanced AI models has sharpened calls across Western countries to build homegrown systems and cut reliance on American technology, Bloomberg reported. On June 13, the US Commerce Department ordered Anthropic to block foreign nationals from accessing Fable 5 and Mythos 5. Within hours, the company disabled both models worldwide for all users, including Americans. It is the first US export control directive aimed at access to a large language model. The directive reaches close allies in the United Kingdom, Canada, Australia and New Zealand, and even covers Anthropic’s own foreign employees.
The government tied its decision to national security. Officials cited a technique to bypass Fable 5’s safeguards that could unlock the cybersecurity capabilities of Mythos, the underlying model. Anthropic disagreed publicly, describing the flaw as a narrow jailbreak limited to one case rather than a universal one, and arguing that similar capabilities could be drawn from other public models, including OpenAI’s GPT-5.5, which face no comparable controls. The company called recalling a model used by hundreds of millions an overreaction. The practical hit was small, since Fable 5 had launched only days earlier and Mythos 5 was limited to select partners, but the signal was large.
Canadian Prime Minister Mark Carney seized on the episode during a visit to Ireland, framing it as a warning about systemic risk and comparing concentrated AI dependence to the bank linkages behind the 2008 financial crisis. He said no one had done anything wrong, but that “we will have done something wrong if we just accept this” without diversifying. Carney added that Canada-US cooperation on AI remains effective and that both sides had flagged risks in Anthropic’s latest models. European figures went further. French politician Jordan Bardella called AI a matter of national sovereignty, while members of the European Parliament Aura Salla and Eva Maydell urged the bloc to stop building on access a foreign government could cut at any time.
The Fallout
The shutdown has accelerated sovereign AI efforts already underway. India has proposed a $5 billion national AI fund, and a British group led by the startup Cosine is rallying companies including BT, HSBC and BAE to build a domestic frontier model. The timing is pointed. The episode lands days before the G7 summit on June 15 to 17, where AI governance tops the agenda and where Anthropic CEO Dario Amodei and OpenAI’s Sam Altman are expected to attend. Carney cautioned that no immediate decisions are likely.
The Deeper Worry
Not everyone framed the ban as government overreach. Some critics argued Anthropic invited the outcome by marketing its models as uniquely dangerous. Cybersecurity researcher Peter Girnus said the company set itself up, writing that “they wrote the legal predicate themselves and called it a brand.” Anthropic has previously challenged Trump administration limits on its technology in court. The broader concern now extends beyond one company. Governments and businesses that depend on a handful of US-based models face the prospect that access could vanish through an export order, an acquisition or a policy shift, which is the core of the sovereignty argument gaining traction this week.
