U.S. Lifts Restrictions on Anthropic's AI Models After Security Enhancements

By Priya Anand

The U.S. government has lifted restrictions on Anthropic's advanced AI models, Fable 5 and Mythos 5, following the implementation of new security measures. This decision comes after the models were previously suspended due to national security concerns.

On June 12, 2026, the U.S. government issued an emergency export control directive requiring Anthropic to suspend access to these models for all foreign nationals, citing a security vulnerability. The concern centered around a method that allowed Fable 5 to read code and fix software flaws, a capability also present in other AI models.

In response, Anthropic collaborated with government agencies and private researchers to enhance the models' safety features. The company developed a new safety classifier that intercepts flagged requests, redirecting them to a less-capable model, Opus 4.8, when sensitive capabilities are probed. This classifier reportedly blocks the identified vulnerability in over 99% of cases.

On June 26, 2026, the Commerce Department approved the limited restoration of Mythos 5 for over 100 U.S. organizations, including government agencies and critical infrastructure defenders. Foreign-national access through these organizations is now permitted.

On July 1, 2026, the Commerce Department lifted export restrictions on Fable 5, allowing global access with certain usage limits for specific user tiers until July 7. This includes re-enabling access on major cloud platforms.