US Lifts Export Curbs on Anthropic’s Mythos and Fable Models

The initial June 12 mandate had classified these AI models as export-restricted technologies, forcing Anthropic to shutter public access because verifying the citizenship of every user proved unmanageable. Under the new agreement, Anthropic has pledged to proactively monitor for security risks and maintain open communication with federal agencies regarding potential malicious activity. These commitments mirror safety policies the company had already adopted voluntarily prior to the government intervention.

Industry observers suggest the reversal was driven by mounting pressure to maintain American competitiveness. As Asian firms, including those developing the Fugu and Tulonfeng models, reach capabilities comparable to Mythos, the Trump administration faced criticism that its restrictions were hindering domestic progress. While the White House recently permitted limited releases of Mythos to pre-approved organizations, the broader policy landscape remains volatile. Analysts like Dean W. Ball have repeatedly questioned the administration’s erratic approach to oversight, leaving major AI labs struggling to predict the regulatory environment for future model launches.