The United States government has authorized artificial intelligence company Anthropic to re-release its Mythos 5 model to a select group of customers and partners, reversing a previous restriction imposed over national security concerns, according to media reports on Friday.
Commerce Secretary Lifts Restrictions
Commerce Secretary Howard Lutnick informed Anthropic that a set of trusted users could once again access the model, as detailed in a letter seen by NBC News. The restoration is expected to cover approximately 100 organizations, including government agencies and private companies, according to people familiar with the matter.
Anthropic's Announcement
Anthropic stated on X: “The government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organisations that operate and defend critical infrastructure.” The company added that since June 12, it had been working closely with the US government to restore access to Claude Mythos 5 and Fable 5.
Model Purpose and Previous Restrictions
Mythos 5 is designed to identify cyber vulnerabilities and support defensive cybersecurity operations. Two weeks ago, Lutnick invoked export control authorities to require Anthropic to suspend access to Mythos 5 and its consumer model, Fable 5, citing national security threats. This action came hours after OpenAI announced that it would release its latest GPT-5.6 models in phases at the federal government's request.
Ongoing Discussions
Anthropic said discussions with government officials would continue over the weekend as it seeks to restore access to Fable 5. According to NBC News, the Trump administration has been developing a framework to test advanced AI models for safety risks while strengthening cybersecurity protections.
Industry Context
OpenAI CEO Sam Altman described the staggered launch of GPT-5.6 as "bad news", noting that the company had initially planned a broader rollout. The situation highlights the increasing government scrutiny of advanced AI models and their potential national security implications.
