The US government just decided who gets access to the world's most powerful AI — and Australia isn't on the list
The US has partially lifted the ban on Anthropic's Mythos 5 but only for hand-picked American organisations.
THE US government has partially lifted its ban on Anthropic's most powerful artificial intelligence model but kept it locked away from the rest of the world, including Australia, in a move that gives Washington veto power over who gets to use frontier AI.
Commerce Secretary Howard Lutnick wrote to Anthropic on Friday confirming roughly 100 "trusted partners" would be permitted to access Claude Mythos 5, the company's most advanced cybersecurity model.
They are all American. Everyone else still needs an export licence.
Fable 5, the public-facing version of the same model that hundreds of millions of people had been using, remains completely banned with no timeline for its return.
On June 9, Anthropic launched both models to widespread acclaim. Three days later Lutnick ordered the company to immediately suspend all access by any foreign national anywhere in the world, including Anthropic's own staff.
Because Anthropic had no way to filter users by nationality it disabled both models for everyone on the planet.
The government cited a reported jailbreak that could bypass Fable 5's safety filters. Anthropic called it "narrow" and argued the same technique worked on OpenAI's GPT-5.5.
The political context is hard to ignore. The Trump administration declared Anthropic a "supply chain risk" earlier this year, a label historically reserved for foreign adversaries. Key Trump advisors have publicly attacked the company as "woke" and "leftist." Anthropic is suing to have the designation reversed.
On the same day Lutnick partially lifted the Mythos ban, OpenAI released its competing GPT-5.6 to government-approved partners.
Lutnick's letter warns all other restrictions remain in effect "until further notice" including the threat of criminal and civil penalties. He reserved "the right to reevaluate and adjust the scope of license requirements should circumstances change."
For anyone outside America who built workflows around these tools and lost access overnight, the message is clear: the US now decides who gets frontier AI and the rest of the world was not consulted.