Anthropic's Mythos and Fable Models Have Their Export Restrictions Lifted

Access Restored to the Most Advanced AI Models

The United States has announced the lifting of export restrictions that had been imposed on Mythos and Fable, two models released by Anthropic and considered among the most sophisticated AI models available today. The company announced it will begin restoring access to both models starting July 1, 2025, bringing an end to a crisis that lasted several weeks.

How the Crisis Began

On June 12, the U.S. government added both models to its list of export-controlled technologies, requiring Anthropic to obtain prior government licensing before making them available to any user outside the United States. Unable to comply with this requirement at scale, the company was forced to suspend public access to both models entirely.

Mythos had been launched in April 2025 for a limited set of organizations amid concerns about its ability to discover and exploit software vulnerabilities. Fable was released to the general public in June, equipped with additional safety layers.

Terms of the Restrictions' Removal

Commerce Secretary Howard Lutnick announced that Anthropic agreed to a set of commitments in exchange for lifting the restrictions, most notably:

• Proactively identifying and addressing security risks associated with the models.
• Ongoing cooperation with the U.S. government in establishing release protocols and standards for Mythos, Fable, and future models.
• Notifying the government immediately upon detecting any harmful or suspicious activity.

Notably, Anthropic had already voluntarily committed to most of these obligations months before the restrictions were ever issued, leading cybersecurity experts to question the true motivations behind the ban.

Were the Restrictions a Political Pressure Tool?

A number of analysts argued that the restriction decision was less a security measure and more a political message. Anthropic executives had previously and publicly criticized the U.S. administration's approach to AI technology, and these analysts believe the ban was a response to that criticism rather than a genuine reaction to security concerns.

Asian Competition Accelerates the Resolution

Domestic political pressure was not the only factor in lifting the restrictions. Asian AI companies began releasing models approaching the capabilities of Mythos, including the Fugu and Tulonfeng models. This placed the U.S. government in a genuine dilemma: either ease restrictions to allow American AI to compete globally, or risk losing ground in the global technology race.

A Situation That Remains Unclear

Last week, Lutnick authorized Mythos to be made available to select customers subject to White House approval, while OpenAI's latest models underwent a similar process. However, the administration's inconsistent approach to shaping AI policy continues to unsettle technology companies and clouds any clear picture of the regulatory framework that will govern future model releases. Prominent analysts, including Dean W. Ball, who recently joined OpenAI, have criticized the June executive order hinting at pre-launch model reviews, calling it a barrier to innovation.

In summary, this crisis exposes the fragility of the United States' AI regulatory framework and highlights the urgent need for clear, consistent policies that balance national security requirements with the imperatives of technological competitiveness.

✦ بقلم فريق دروب أيديا