US Government Approves Anthropic to Release Claude Mythos to Trusted Partners

Anthropic’s most powerful — and most tightly restricted — AI model just got a partial green light from Washington. After a tense few weeks involving a national security scare, Claude Mythos 5 is heading back to a select group of vetted US organizations.

What Happened

The US government has allowed Anthropic to release its Claude Mythos 5 AI model to some “trusted” US organizations, partially reversing an order issued two weeks earlier that suspended access over national security risks. More than 100 companies and institutions, including many Fortune 500 firms, will now have access to Mythos 5.

Anthropic confirmed it received notice that Mythos 5, its strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers, and said it is working to restore their access as quickly as possible.

Why Access Was Suspended in the First Place

Anthropic announced Claude Mythos Preview on April 7, 2026, under Project Glasswing, with the initial cohort capped at approximately 200 organizations including US government entities. A mid-June 2026 emergency directive from the US government temporarily suspended access, citing national security concerns tied to a reported jailbreak vulnerability.

Fable 5, the public-facing version of the same underlying model, is prevented from responding to certain cybersecurity and biology-related queries, with those queries routed instead through Claude Opus 4.8. The Commerce Department’s letter does not mention any change to restrictions on Fable 5 itself.

What Mythos Actually Does

Claude Mythos Preview is a general-purpose, unreleased frontier model that demonstrates AI has reached a level of coding capability where it can surpass all but the most skilled humans at finding and exploiting software vulnerabilities, having already found thousands of high-severity flaws across every major operating system and web browser.

We covered this exact trajectory closely as it unfolded — from the original Project Glasswing announcement and its launch partner list to the Akrites Security Body that Anthropic, Google, Microsoft, and OpenAI jointly formed in response to AI’s accelerating vulnerability-discovery speed.

The Bigger Picture: Export Controls on AI

An export license will no longer be needed for Mythos 5 access among trusted companies and their non-US-citizen employees, though licensing restrictions remain for companies outside the approved list. Both Anthropic and OpenAI plan to go public, but Anthropic’s relationship with the US government has been particularly rocky. The episode also affected OpenAI, which limited the release of its own capable model, GPT-5.6, under similar pressure from the administration.

What This Means Going Forward

This is a developing regulatory situation, and details can shift quickly given the sensitivity involved. For the most current, authoritative status on Mythos and Fable access, check Anthropic’s official Project Glasswing page and its news announcements directly — this is a fast-moving story and Anthropic’s own channels will reflect any further changes faster than third-party coverage.