US lifts export controls on Anthropic's Fable 5 model after two and a half weeks

The US Department of Commerce lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5 models on June 30, 2026, roughly two and a half weeks after imposing them on June 12. Anthropic said Fable 5 returned to users on July 1 across Claude.ai, the Claude Platform, Claude Code and Claude Cowork.

The reversal ends a standoff that had taken both models offline worldwide. The original order required Anthropic to cut off Fable 5 and Mythos 5 access for any foreign national – including its own non-citizen staff – after the government cited a discovered method for bypassing the models’ safeguards. Anthropic said it had no reliable way to verify a user’s nationality in real time and suspended both models globally rather than comply selectively.

The episode showed how a single export directive can pull a widely used AI model offline for every customer, not just those in the targeted jurisdictions. Mythos 5, the more tightly controlled sibling model, had partial access restored for select US organizations on June 26, ahead of the broader lift.

Anthropic paired the redeployment with new safeguards. It said a new safety classifier now blocks the specific jailbreak technique in more than 99 percent of cases, and it is proposing an industry-wide jailbreak-severity scoring framework with Amazon, Microsoft, Google and other partners. Anthropic is also offering Fable 5 at up to 50 percent of weekly usage limits on paid plans through July 7.

The 99 percent block rate is Anthropic’s own figure and has not been independently verified; a classifier that stops most attempts of the specific technique is not proof it stops every variant of the bypass that triggered the controls. Whether a severity-scoring standard gains traction with rivals – or stays an Anthropic proposal – will determine if this becomes shared industry practice or a one-company response to a one-company incident.