
Anthropic Mythos 5 limited release to US defenders after government detente
Published by AINave Editorial • Reviewed by Ramit
The Trump administration has allowed Anthropic to redeploy its strongest cybersecurity model, Mythos 5, to a set of over 100 US government agencies and private organizations for defensive cyber purposes. The move ends a two-week standoff that began when Commerce Secretary Howard Lutnick invoked export control authorities to shut off access to both Mythos 5 and Fable 5, citing national security concerns after guardrails were allegedly bypassed by security researchers.
What happened
On June 26, Commerce Secretary Howard Lutnick sent a letter to Anthropic's chief compute officer Tom Brown stating that "appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model" TechCrunch. The directive allows non-American employees at approved organizations and Anthropic itself to access Mythos 5, reversing the original ban that barred all foreign nationals CNBC.
Anthropic confirmed the development on X, stating that Mythos 5 "can be redeployed to a set of US organizations that operate and defend critical infrastructure" NBC News. The company said it is restoring access quickly and continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again The Verge.
Fable 5, the consumer-facing model built on the same technical foundation as Mythos 5 but with stronger guardrails, was not addressed in the directive and remains restricted TechCrunch.
Why AI builders should care
For teams building AI-powered cybersecurity products or defending critical infrastructure, Mythos 5 represents a significant capability. An early version of Mythos 5 was able to find thousands of new cyber vulnerabilities and bugs NBC News. Access is now limited to approved organizations, but the episode signals how the US government will handle future powerful model releases.
The government did not lift the export control directive entirely. Instead, it created a limited preview exception similar to the one applied to OpenAI's GPT-5.6 family, announced the same day The Verge. Both Anthropic and OpenAI have stated they hope for broader availability in the coming weeks, but the process remains ad-hoc and subject to ongoing negotiations NBC News.
Practical implications
If your organization operates or defends critical infrastructure, you may be eligible for Mythos 5 access through the approved list. The list includes government agencies and private partners such as Cisco and JPMorgan Chase, which were part of Anthropic's Project Glasswing before the ban NBC News.
For builders outside the approved set, Fable 5 remains unavailable. Anthropic is continuing discussions with the government to restore general access, but no timeline has been provided The Verge. The episode also highlights that future model releases from leading labs may face similar case-by-case government review, especially for models with dual-use capabilities.
Caveats
The government retains the right to reevaluate and adjust the scope of license requirements on Mythos 5 and Fable 5 should circumstances change The Verge. All other requirements of the June 12 letter remain in effect until further notice. The situation is time-sensitive and reflects ongoing negotiations between Anthropic and the Trump administration.
FAQs
What is Anthropic Mythos 5 and what makes it different from other AI models?
Mythos 5 is Anthropic's strongest cybersecurity model, designed for defensive cyber use with guardrails and risk controls TechCrunch. It is distinct from Fable 5, which is a consumer-facing variant with stronger guardrails but built on the same technical foundation NBC News.
Who can access Mythos 5 and under what conditions?
Access is limited to over 100 US government agencies and private organizations approved by the Commerce Department CNBC. Non-US nationals within those approved organizations, including Anthropic's own non-American employees, are now allowed to access the model TechCrunch. The export control directive remains in effect for all others.
Why did the U.S. government lift some restrictions on Mythos 5?
Commerce Secretary Howard Lutnick determined that Anthropic had worked with the government to address risks associated with the model and that appropriate safeguards were in place for trusted partners The Verge. The limited redeployment is intended to support defensive cyber operations for critical infrastructure NBC News.
What is Fable 5 and is it included in the current access?
Fable 5 is the consumer-facing version of Mythos 5 with stronger guardrails that was previously available to the general public NBC News. It was not addressed in the June 26 directive and remains restricted TechCrunch. Anthropic is continuing negotiations to restore general access.
Sources
- Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies
- Trump admin allows Anthropic to release Mythos AI model to some companies, government agencies: Reports
- Would You Eat An AI-Designed Fast Food Burger? Researchers Say You Might Not Know The Difference
- U.S. government gives Anthropic green light for limited re-release of Mythos 5
- Anthropic’s Mythos 5 is back
- The US government has allowed Anthropic to release its powerful ...
- Trump Administration Allows Anthropic to Release Mythos to ...
- Trump Admin releases Anthropic Mythos to be used by more than ...
- Trump admin allows Anthropic to release Mythos AI model to some companies, government agencies: Reports
- U.S. Loosens Restrictions on Anthropic’s Mythos A.I. Model
- Claude Mythos 5 Cleared for 100 US Institutions: Will Fable 5 Follow?
- US government allows Anthropic limited release of AI model that sparked cybersecurity concerns
- Federal government permits release of Anthropic’s Mythos model to select companies
- Trump Administration Unveils Anthropic Mythos for Over 100...
- Anthropic Releases Powerful Mythos Model to Select US Companies
- US releases powerful Anthropic model Mythos to some US... | Semafor
- Anthropic moves toward deal with U.S. to lift curbs on AI models