U.S. Approves Limited Rollout of Anthropic's Mythos 5 AI Model

The U.S. government has granted conditional approval for a limited rollout of Anthropic's advanced artificial intelligence model, Mythos 5. This decision allows a select group of trusted American cybersecurity firms and infrastructure providers to regain access after the model's suspension earlier this month due to national security concerns.

Anthropic announced on Friday that it received official notification to restore access to Mythos 5 for specific organizations. The company stated its ongoing commitment to discussions with U.S. authorities to expand access further and eventually make its less powerful Fable 5 model available to the public again. This partial reversal follows restrictions imposed on June 12, when the U.S. government ordered Anthropic to disable both Mythos 5 and Fable after discovering vulnerabilities in the safeguards designed to prevent misuse.

“

I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model.

U.S. Commerce Secretary Howard Lutnick informed Anthropic in a letter that the company had successfully addressed the risks associated with the advanced AI models. "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model," Lutnick wrote. Anthropic confirmed this development, stating, "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."

The company is now working to provision access for the approved providers as quickly as possible. Discussions between Anthropic and the government are expected to continue over the weekend, with the primary goal of restoring general access to Fable 5. The restrictions on Anthropic occurred amidst increasing concerns about the potential exploitation of advanced AI systems by malicious actors to identify and exploit cybersecurity vulnerabilities. Rival OpenAI also recently launched its GPT-5.6 model with restricted access pending government validation, reflecting a broader trend of increased scrutiny and regulation of advanced AI technologies.

“

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.