US Blocks G7 Access to Anthropic's Fable 5 and Mythos 5

The United States has refused to grant G7 allies exceptions to its export ban on Anthropic's most advanced AI models, leaving Fable 5 and Mythos 5 unavailable worldwide.

The US Department of Commerce imposed an export ban on the models for national security reasons. The ban targets foreign nationals and has been in effect for five days, erupting during the G7 summit at Évian-les-Bains. Because Anthropic cannot verify user nationality in real time, the company disabled access to Fable 5 and Mythos 5 for all users globally, including in France and for its own employees, according to multiple reports.

The decision stems from a reported jailbreak vulnerability that Washington says could allow the models to identify software weaknesses for potential hacker use. Anthropic disputes the severity, calling the flaw minor and arguing the response is disproportionate for a commercial model used by millions. The company states its safeguards were previously tested with government agencies.

Britain's request for an exception was rejected as illogical. British Prime Minister Keir Starmer asked President Donald Trump for relief for the UK on the summit sidelines. An administration official told the New York Post that granting exceptions to allies would be perfectly illogical, and the US has no intention of reversing course. Reuters reports G7 leaders discussed a broader trusted partners framework for selective AI access, with talks involving Commerce Secretary Howard Lutnick.

Forbes reports that Lutnick sent a letter to Anthropic CEO Dario Amodei outlining the export control directive. Anthropic issued an official statement on the order. The standoff continued even as Amodei addressed G7 leaders at a June 17 lunch, urging them not to divide over AI while Trump sat at the table. Trump said negotiations with Anthropic were going well.

The dispute builds on prior tensions between Anthropic and the US government. Earlier this year the company was classified as a national security risk after refusing to supply models for surveillance and autonomous weapons. Anthropic has also sued the United States over a related Claude ban. Amazon, the company's largest investor, reportedly originated the vulnerability alert that triggered the current action.

European users now face the reality that a foreign AI tool can be withdrawn overnight by another government's decision. Other Anthropic models remain accessible. The fear that these systems could reach the wrong hands persists.