The Commerce Department lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5 on Tuesday, ending an 18-day standoff that had cut off the company’s two most capable models from every foreign national on the planet, including Anthropic’s own non-citizen employees.

The reversal arrives with an unusual level of choreography. Commerce Secretary Howard Lutnick announced on X that his department had “worked closely with Anthropic to analyze and approve Fable 5,” and addressed the formal letter to co-founder Tom Brown, who had taken over the negotiations from chief executive Dario Amodei. Fable 5 returns to global users on July 1 across Claude.ai, the Claude Platform, and Claude Code, with weekly usage capped at 50 percent through July 7. Access on Amazon Web Services, Google Cloud, and Microsoft Foundry is expected to follow.

The June 12 directive was triggered by a jailbreak technique demonstrated by Amazon researchers, which prompted Fable 5 to identify software vulnerabilities in a codebase. Anthropic’s fix is a new safety classifier that it says blocks the disclosed method in more than 99 percent of attempts, re-routing flagged requests to Claude Opus 4.8. Mythos 5 had already been partially restored on June 26 for roughly 100 U.S. organisations defending critical infrastructure under Anthropic’s Glasswing program.

What’s notable is how quickly a national-security cutoff of a frontier model became a bilateral technical negotiation with a fixed exit ramp. Anthropic is now proposing an industry-wide jailbreak severity framework with Amazon, Microsoft, and Google. The episode looks less like the CFIUS-style enforcement it mimicked, and more like the beginning of a private-sector-authored compliance regime the government agrees to co-sign.

Sources