White House orders Anthropic to restrict AI exports

What happened

The White House ordered Anthropic to restrict export of its AI models Fable and Mythos to anyone outside the United States and foreign nationals inside the country, citing unspecified national security concerns. Anthropic pulled both models within roughly 90 minutes of notification. The ban followed two events: Anthropic giving a South Korean telecom (reportedly SK Telecom) access to Mythos through its partner program, and Amazon CEO Andy Jassy alerting the administration after Amazon researchers found a way around Fable 5's safeguards. Anthropic disputes the characterization, calling it a narrow, already-patched issue.

Why it matters

This is the first real test of whether the U.S. government can use export controls to contain frontier AI, echoing decades of attempts to restrict encryption and spyware with uneven results. The U.S. has a poor track record: it criminally investigated PGP's creator in the 1990s over encryption export but ultimately lost, leading to the widespread adoption of end-to-end encryption by billions of Signal and WhatsApp users. Later, the Wassenaar Arrangement treaty on spyware exports has been weakened by countries that don't adhere (including Israel) and lax enforcement by Europe, home to many spyware makers. How this standoff resolves could shape the rulebook other AI labs must follow.

What to watch

The impasse remains unresolved. There is a reasonable chance the Trump administration will lift the restriction to keep American AI companies competitive worldwide, which would tacitly acknowledge that AI labs elsewhere, including in China, will likely reach similar capabilities regardless of restrictions. Alternatively, American AI companies could face a requirement to get government approval before serving foreign customers—a compliance burden that would dent their bottom line.