If you were a developer outside the United States on June 12, your Fable 5 access vanished mid-workflow. No warning. No grace period. The Trump administration imposed export controls on Anthropic‘s two most powerful commercial AI models, ordering the company to cut off every foreign national — including its own foreign-national employees. Eighteen days later, Commerce Secretary Howard Lutnick lifted those restrictions entirely. The price: Anthropic now operates under formal, ongoing government oversight of its AI releases.
What Happened — and How Fast
Three weeks turned Anthropic’s flagship models into a geopolitical bargaining chip.
The timeline tells the story.
- June 12: Commerce ordered a full shutdown of Fable 5 and Mythos 5 access for any foreign national, anywhere in the world
- June 26: Partial easing granted Mythos 5 access to roughly 100 vetted U.S. critical-infrastructure organizations only
- June 30: Lutnick confirmed full withdrawal, writing that “a license is no longer required for the export, reexport, or in-country transfer of the Mythos or Fable models”
- Fable 5 restores globally via Claude.ai, Claude Code, AWS, Google Cloud, and Microsoft Foundry
- Mythos 5 remains restricted to select trusted U.S. partners under Anthropic’s Glasswing program, with expansion planned
Eighteen days is not a thorough national-security review. It’s a negotiation. Lutnick letter to Anthropic reserves the right to re-impose licensing if commitments slip “We have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America’s leadership in AI.” The leash stays attached.
The Deal Anthropic Made
Anthropic traded a degree of autonomy for restored access, agreeing to a co-governance framework that could reshape how every frontier lab operates.
Anthropic committed to:
- Proactively detecting security risks
- Sharing jailbreak techniques with federal testers
- Providing pre-release access to future national-security-relevant models
- Reporting malicious activity directly to Commerce and the National Cyber Director’s office
Commerce’s Center for AI Standards and Innovation reportedly called the updated safeguards “extraordinarily strong,” according to Bank Info Security. Anthropic agreed to “inform the U.S. government of any malicious activity” involving its systems, according to Reuters.
If your enterprise workflows depend on frontier AI tools, this episode is worth a hard look at your contingency planning. Washington demonstrated it can shut down a leading commercial AI platform globally with a single directive — no compensation, roughly two weeks of disruption. Think of it as the AI-era equivalent of a major cloud region going offline, except the outage was deliberate and resolved only after the provider agreed to permanent audit rights.
The export controls are lifted. The precedent they set — that frontier AI labs will cooperate with Washington or lose global distribution — remains firmly in place. Anthropic’s arrangement around pre-release access, formal standards cooperation, and jailbreak sharing may well become the template Washington applies to every other major lab next.
