Swansenreport

Anthropic launched its most powerful publicly available AI model, Claude Fable 5, on June 9. Paying subscribers got a two-week free window to try it before it moved behind a usage-based paywall. What followed was a lesson in what happens when Washington decides it doesn’t like your product launch.

Three days in, the feds pulled it.

The Commerce Department issued an export control directive requiring Anthropic to suspend all access to Fable 5 and Mythos 5 by any foreign national — anywhere in the world, including Anthropic’s own foreign-born employees. The directive arrived at 5:21 p.m. ET on June 12. The letter offered no specific national security rationale. Anthropic’s read: the government had learned of a jailbreak technique involving a small number of previously known, minor vulnerabilities that other publicly available models could also expose. Anthropic + 2

It was the first time Commerce had ever used that particular provision of the Export Control Reform Act against a commercial AI model. Cyber Security News

Anthropic didn’t try to sort users by passport. It shut the thing off globally. The company had filed a confidential IPO prospectus weeks earlier, disclosing a $47 billion revenue run rate and a $965 billion valuation. A government kill-switch dropping just before a potential public offering was not a great look. Some analysts said it would make investors question whether Anthropic could stay at the frontier if Washington kept singling out its models. QuartzFortune

Access came back around June 18. But the promised two-week free window had already burned down to maybe four or five usable days. Today, June 23, Fable 5 is off the subscription menu entirely. Using it now draws from usage credits at $10 per million input tokens, $50 per million output. Claude Fast

The backstory makes it weirder. Earlier this year, the Pentagon declared Anthropic a supply chain risk — a label usually reserved for foreign adversaries — after contract talks broke down. That lawsuit is still active. In February, Trump ordered federal agencies to stop using Claude after Anthropic balked at contract terms that would have allowed its models to be used for autonomous weapons systems. FortuneFortune

Anthropic pushed back hard on the jailbreak rationale. The company said the model had been red-teamed for thousands of hours before launch by the U.S. government, the U.K. AI Security Institute, and third-party organizations — none of which found a universal jailbreak. Anthropic

Fable 5 is available today, if you want to pay for it. Whether the government leaves it that way is the open question.