The United States has lifted export restrictions on Anthropic's Mythos and Fable AI models, clearing the way for the company to restore global access after weeks of regulatory uncertainty.

The move allows Anthropic to resume international availability of its flagship AI models from July 1, following an agreement with the US government on security and safety measures.

Global Access Returns After Weeks of Restrictions

In June, the US government classified Mythos and Fable as export-controlled technologies, requiring Anthropic to obtain licences before making them available outside the United States.

The restrictions proved difficult to implement at scale, prompting Anthropic to suspend public access to both models.

Following discussions with US officials, the government has now removed the licensing requirement, allowing the company to begin restoring access worldwide.

Anthropic Agrees to New Security Commitments

As part of the agreement, Anthropic will work closely with the US government to strengthen safeguards around its frontier AI models.

According to the US Department of Commerce, the company has agreed to:

  • Proactively detect and respond to AI-related security risks.
  • Collaborate on future AI safety protocols and model releases.
  • Report malicious use of its models to US authorities.

Anthropic had previously introduced many of these safety commitments voluntarily before the export restrictions were imposed.

Competition in the AI Race

The policy reversal comes as competition in the global AI market continues to intensify.

Recent launches from Asian AI companies, including Japan's Sakana AI with its Fugu model and China's 360 with Tulongfeng, have increased pressure on US policymakers to ensure American AI companies remain competitive internationally.

The restrictions had temporarily limited Anthropic's ability to compete in overseas markets while rival models continued expanding.

What It Means

The decision restores international access to some of the world's most advanced AI models while signalling a more collaborative approach between AI developers and US regulators.

However, the episode has also highlighted ongoing uncertainty around US AI policy, with frontier AI companies still seeking clearer rules on how future advanced models will be reviewed and released.