The Trump administration continues to restrict Anthropic's most powerful model, Mythos, making it inaccessible to overseas organizations.

```

While the U.S. government has lifted export controls on Anthropic PBC’s latest artificial intelligence model, its more powerful Mythos 5 model remains restricted for use by a handful of American institutions, leaving global users—including foreign governments, enterprises, and financial institutions—still facing uncertainty.

According to Bloomberg, Anthropic stated that global users can access the Fable 5 model, which is similar to Mythos and intended for a broader audience, starting Wednesday. However, negotiations to expand access to Mythos 5 inside and outside the U.S. through the “Glasswing Initiative” are still ongoing, with the U.S. government to ultimately decide which institutions regain access.

An Anthropic spokesperson said the company is restoring access for existing American partners under the previous Glasswing Initiative, but there is no timeline for when international partners will regain access.

The lack of transparency from Washington regarding which institutions may use these models highlights the extent to which the digital security of many U.S. allies depends on American policy. At the same time, it also makes Anthropic a “test case” for the U.S. regulation of cutting-edge AI models.

Export Controls Lifted, but Mythos Access Still Restricted

Mythos is extremely powerful in uncovering cybersecurity vulnerabilities, and Anthropic previously limited access to a vetted group of institutions to allow vulnerabilities to be discovered and patched before being exploited by bad actors. Anthropic says Fable 5 is just as capable, but with built-in security protections preventing the model from being used to expose cybersecurity flaws.

Anthropic has never released Mythos to the public. This past April, the company opened access through the Glasswing Initiative to an initial group of 50 institutions, expanding by June to 150 institutions in 15 countries.

This progress was interrupted on June 12 by export controls from the U.S. Department of Commerce, which required Anthropic to obtain U.S. government approval before allowing any foreign national (regardless of location) to access Fable or Mythos.

According to Anthropic, this control order was issued after reports that Amazon security researchers had found ways to bypass Fable 5’s security features, making it possible for the model to be used to mine cybersecurity vulnerabilities.

Last week, the U.S. eased some restrictions on Mythos, and on Tuesday formally lifted export controls. According to people familiar with the matter, this shift came after Anthropic added a new classification (i.e., safety filtering mechanism) to Fable 5.

U.S. Approval Process Opaque, Institutions Treated Unequally Across Countries

During earlier stages of access, approval for institutions in different countries was not uniform. The UK’s AI Security Institute was among the few international bodies allowed to test Mythos early on.

The EU’s cybersecurity agency ENISA, after weeks of negotiation, was also arranged for access. Japanese banking institutions were likewise included on the Mythos testing list, but European banking institutions were not granted the same permissions.

This uneven access pattern further reflects the dominant role of U.S. policy in determining who can use this cutting-edge tool, and continues to leave overseas users—including financial institutions—dependent on the pace of Washington’s approvals for their cybersecurity capabilities.

Risk DisclaimerThere are risks in the market, and investments should be made with caution. This article does not constitute personal investment advice and does not consider the individual investment goals, financial situation, or needs of any particular user. Users should consider whether any opinions, views, or conclusions in this article fit their own circumstances. Investment based on this article is at your own risk. ```