The United States has lifted a licensing requirement that required Anthropic to obtain government approval before exporting its Mythos and Fable AI models overseas, a restriction that had effectively blocked public access to what are widely regarded as the most advanced AI models released so far.

Anthropic said it will begin restoring access to the models on Wednesday, July 1.

On June 12, the U.S. government placed Anthropic’s Mythos and Fable models on its export control list, requiring the company to obtain a license before making them available to foreign nationals. Because complying with the rule at scale was impractical, Anthropic suspended public access to the models.

Following weeks of discussions, U.S. Commerce Secretary Howard Lutnick said Anthropic had agreed to strengthen safeguards by proactively identifying and addressing security risks tied to the models, working closely with the U.S. government on protocols, standards and future releases, and reporting any malicious activity involving the technology.

Anthropic had publicly committed to many of those safeguards months before the export restrictions were introduced, leading some cybersecurity experts to question the need for the measures.

They argued the restrictions appeared less focused on national security and more like a source of leverage, suggesting the Trump administration may have been responding to public criticism from Anthropic executives over how the government, and the president’s political opponents—could use advanced AI technology.

Mythos was initially rolled out to a limited group of organizations in April to address concerns about its ability to identify and exploit software vulnerabilities. A more restricted version, Fable, was released to the public in June with additional safety guardrails.

However, as Asian AI companies began unveiling models with capabilities approaching those of Mythos—including Fugu and Tulongfeng—the U.S. government faced growing pressure to relax the restrictions on Anthropic to help ensure American AI firms remained competitive in the global market.

Last week, Commerce Secretary Howard Lutnick approved the release of Mythos to a select group of White House-authorized customers. Similarly, OpenAI’s latest models were made available only to organizations approved by the Trump administration, rather than being released publicly.

The Trump administration’s shifting approach to AI regulation has created uncertainty across the industry, leaving companies with limited clarity over the rules that will govern future AI model releases.