US Allows Anthropic To Restore Mythos 5 AI Model Access To Select Trusted Partners

The US government has authorized artificial intelligence startup Anthropic to resume access to its powerful Claude Mythos 5 model for select companies and organizations, reversing a sweeping export freeze implemented earlier this month amid national security concerns.

In a letter sent to Anthropic on Friday, Commerce Secretary Howard Lutnick confirmed that the company had successfully addressed immediate security concerns. While the decision allows a restricted pool of trusted infrastructure providers and cyberdefenders to resume using Anthropic's most advanced model, a broader suspension remains in place for its secondary model, Fable.

Limited Re-deployment and Guardrails

The regulatory shift comes after Anthropic spent weeks negotiating with federal officials to implement appropriate safeguards. The Commerce Department's decision establishes a highly restricted path forward, focusing access specifically on critical infrastructure defense. Anthropic Sends Senior Staff to Washington to Resolve White House Dispute Over Export Restrictions on Mythos 5 and Fable 5 AI Models.

“We received notice from the U.S. government that Mythos 5, our strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers,” Anthropic said in a statement. The company added that it is working to provision access for approved providers as quickly as possible.

However, the federal authorization remains partial. Secretary Lutnick’s letter did not extend permission for Anthropic to reactivate Claude Fable 5, a less powerful variant of the Mythos architecture originally designed for broader enterprise deployment. Anthropic stated that it continues to work with the government to make Fable 5 available for general use again, and sources familiar with the matter indicate discussions are expected to continue over the weekend. US Government Orders Anthropic To Suspend Access to Claude Fable 5, Mythos 5 Over Export Control Directive.

National Security Fears and Technical Power

The initial export block, enacted on June 12, forced Anthropic to abruptly disable customer access to both models. The federal directive was designed to prevent foreign nationals - including some of Anthropic’s own international staff - from utilizing the technology.

The government’s intervention followed mounting anxiety among national security experts regarding the unprecedented capabilities of the Mythos architecture. Unlike standard consumer AI, Mythos was engineered with advanced reasoning capabilities tailored for software security.

During internal testing, the model demonstrated the ability to autonomously scan complex codebases, discovering long-hidden vulnerabilities in hours that had eluded human analysts for decades. Regulators feared that if the technology fell into the hands of foreign state actors or sophisticated hacking groups, it could significantly accelerate automated cyber warfare and exploit development.

Regulatory Tension and Broader Friction

The swift block and subsequent partial reversal underscore an ongoing struggle within Washington to balance technological dominance with national security. While the current administration has generally favored a hands-off approach to foster innovation and outpace global competitors like China, the sheer capability of the Mythos model forced an aggressive pivot.

This export dispute is not the first time Anthropic has clashed with federal oversight. Earlier this year, the U.S. government designated the startup a "supply chain risk," effectively blacklisting it following an internal disagreement regarding the permissible scope of military applications for Anthropic's products. Anthropic filed a lawsuit challenging that designation and has secured at least one favorable early ruling in the ongoing legal proceedings.

For now, the partial resolution for Mythos 5 represents a critical test case. How Anthropic handles the restricted rollout could dictate how future frontier AI architectures are monitored, licensed, and deployed within the United States.