Updated · Peterson Institute for International Economics · Jul 2
US Commerce Department Curbs Anthropic's 2 Top AI Models, Restoring Fable Access on July 1
Updated
Updated · Peterson Institute for International Economics · Jul 2
US Commerce Department Curbs Anthropic's 2 Top AI Models, Restoring Fable Access on July 1
3 articles · Updated · Peterson Institute for International Economics · Jul 2
Summary
June 12 export controls forced Anthropic to block non-US access to its Mythos and Fable models, and the company briefly took both offline for all users because it could not verify citizenship.
US officials acted over fears frontier AI could aid adversaries or leak strategic capabilities, but the legal basis for controlling cloud-based model access remains contested.
July 1 brought only a partial rollback: Fable returned for international users with tighter guardrails, while Mythos remains limited to US institutions and foreign nationals at US institutions were later exempted.
Emmanuel Macron called the move "nationalist" at the G7 and warned allies would shun US models if Washington could "turn off the switch" without notice.
The episode is expected to push some overseas users toward Chinese open-weight models, sharpening concerns that AI access is becoming a geopolitical supply-chain chokepoint.
Did the US effort to control AI just hand the global market over to its Chinese rivals?
After the sudden AI shutdown, is the dream of open access to frontier technology now dead?
Anthropic’s Claude Fable 5 Crisis: How a 3-Week US Export Ban Reshaped AI Safety, Regulation, and Industry Standards in 2026
Overview
In June 2026, Amazon researchers discovered a way to bypass safety protocols in Anthropic’s Claude Fable 5 AI model, raising concerns about its potential misuse. Because Claude Fable 5 had not completed the required voluntary review for frontier models, the U.S. government quickly imposed strict export controls as an urgent response. This led to a complete restriction of Claude Fable 5 and only a partial reopening of Mythos 5, which was seen as less risky. The crisis highlighted the need for rapid safeguards, government oversight, and industry collaboration to balance AI innovation with security.