Trump administration clears Anthropic’s Mythos 5 for use by over 100 US agencies and companies

The Trump administration has reversed course on its sweeping ban of Anthropic’s most advanced cybersecurity models, allowing the company to redeploy Claude Mythos 5 to more than 100 designated U.S. government agencies and critical infrastructure companies. The decision, communicated in a letter from Commerce Secretary Howard Lutnick to Anthropic’s chief compute officer Tom Brown on Friday, marks a significant softening of restrictions imposed just two weeks ago.

Background of the ban and partial reversal

On June 12, the administration ordered Anthropic to pull both Mythos 5 and Fable 5 from the market after security researchers demonstrated that the models’ guardrails could be bypassed with relative ease. The ban also prohibited non-American employees — including those at Anthropic — from accessing the models, a restriction that drew criticism from industry groups concerned about hampering U.S. competitiveness in AI security.

Under the new directive, Mythos 5 can now be accessed by a curated list of organizations that operate and defend critical infrastructure. The list includes not only U.S. government agencies but also private sector companies, and notably extends access to non-American employees working at those organizations. Anthropic’s own non-American staff, who were initially barred, are also included in the latest authorization.

“I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” Lutnick wrote in the letter, as reported by Semafor and Reuters. The directive does not address Fable 5, a version of the model that was released more broadly days before the ban and was considered to have stronger protective measures.

Anthropic’s response and next steps

Anthropic acknowledged the development in a post on X on Friday, stating: “Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure.”

The company said it is “restoring access for these organizations quickly” and continues to work with the government to expand access to Mythos 5 and to make Fable 5 available for general use again. Anthropic did not immediately respond to a request for comment on the timeline for broader availability.

Why this matters

The partial reversal reflects the delicate balance the administration is trying to strike between national security concerns and the need to maintain U.S. leadership in AI. Mythos 5 and Fable 5 are among the most powerful cybersecurity models ever developed, capable of autonomously identifying and patching vulnerabilities. Restricting access too broadly risked leaving critical infrastructure exposed, while unrestricted access raised fears of misuse by adversaries.

The decision also signals that the administration is willing to tailor restrictions based on trust and verification, rather than imposing blanket bans. This case-by-case approach may become a template for how other advanced AI models are regulated in the future.

Conclusion

The Trump administration’s decision to allow over 100 U.S. agencies and companies to access Anthropic’s Mythos 5 marks a calibrated step toward balancing AI safety with operational necessity. While Fable 5 remains unavailable for general use, the move provides immediate cybersecurity support to key organizations and sets a precedent for future AI governance. The situation remains fluid, and further developments are expected as Anthropic continues discussions with the government.

FAQs

Q1: What is Anthropic Mythos 5?
Mythos 5 is Anthropic’s most advanced cybersecurity AI model, designed to autonomously identify and patch vulnerabilities in software and network systems. It is part of the Claude family of models.

Q2: Why was Mythos 5 banned?
The model was banned on June 12 after security researchers demonstrated that its built-in guardrails could be bypassed, raising concerns about potential misuse by malicious actors or foreign adversaries.

Q3: Who can now access Mythos 5?
Access is granted to more than 100 specific U.S. government agencies and companies that operate and defend critical infrastructure. Non-American employees at those organizations, as well as Anthropic’s own non-American staff, are also permitted to use the model under the new directive.