The Trump administration on June 12 issued an export control directive ordering Anthropic to suspend all foreign national access to its two most capable artificial intelligence (AI) models, Claude Fable 5 and Mythos 5. Commerce Secretary Howard Lutnick delivered the order to Anthropic CEO Dario Amodei at 5:21 p.m. Eastern, and Anthropic pulled both models offline for all users worldwide within hours.
David Sacks, the administration’s AI czar and co-chair of the President’s Council of Advisors on Science and Technology (PCAST), said on X the government warned Amodei that a trusted partner had identified a jailbreak of Fable 5’s guardrails granting access to the unrestricted cyber capabilities of Mythos, the model it is built on. Sacks said Amodei declined to fix the bypass or pull the model. “The ball is in Anthropic’s court,” he wrote.
The directive marks the first known application of U.S. export control law to a deployed commercial AI model rather than to semiconductor hardware. The order’s nationality-based access restriction made selective compliance technically impossible, forcing a complete global shutdown.
Semafor reported the action was partly triggered by Amazon CEO Andy Jassy alerting White House officials to security concerns and suspicions that a China-linked group had accessed Mythos.
Anthropic said the government provided only verbal evidence of a “potential narrow, non-universal jailbreak” and disputed that the finding warranted a recall, stating the same capability exists on competing models, including OpenAI’s GPT-5.5.
Pentagon Chief Information Officer Kirsten Davies posted on X on June 13 that “some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation. America First. Always.”
Senior Anthropic technical staff flew to Washington on June 14 and met with White House officials, Axios reported. Administration officials said the company had not previously engaged in a serious manner. Sources from both sides said they are eager to resolve the dispute.
