The US government has partially walked back its aggressive block on Anthropic’s flagship artificial intelligence models, granting authorization for a select group of American cybersecurity firms and critical infrastructure providers to deploy Claude Mythos 5. 

The surprise de-escalation comes exactly two weeks after the Trump administration abruptly invoked export control authorities to force Anthropic to freeze access to its two most advanced models, Mythos 5 and Fable 5, citing severe national security and jailbreak concerns. Commerce Secretary Howard Lutnick informed the company via a private letter that intensive, daily negotiations had cleared the path for a controlled deployment under a new, fast-forming regulatory regime. 

“I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” Lutnick wrote to Anthropic’s Chief Compute Officer Tom Brown on Friday afternoon, noting "significant progress" in bilateral discussions. 

The initial re-release will bypass standard export license requirements for roughly 100 heavily vetted organizations—primarily members of Anthropic’s defensive security initiative, Project Glasswing—allowing them and their foreign national employees immediate access to the system. 

The initial June 12 shutdown sent shockwaves through the tech sector, coming just hours after Anthropic officially launched the Mythos-class models. During internal trials, Mythos Preview demonstrated unprecedented, autonomous multi-step exploit capabilities, including uncovering a 27-year-old vulnerability in OpenBSD’s TCP stack and engineering functional exploits for Mozilla’s Firefox browser at a rate 90 times faster than prior generations. Fearing those exact capabilities could be weaponized by foreign adversaries like China or Russia to compromise global banking and infrastructure networks, Washington stepped in. 

The administration's quick U-turn highlights the high-stakes balancing act between national security and keeping American AI labs ahead in a fierce global race. The decision notably coincided with OpenAI’s Friday release of its own latest model, GPT-5.6, which is also utilizing a restricted, government-validated rollout. 

“In just two weeks, we have worked diligently to ensure America remains the global leader in AI while safeguarding our security,” Commerce Department spokesperson Benno Kass said. 

While the letter marks an engineering win for Anthropic, the broader commercial landscape remains deeply fractured. The government’s directive is completely silent on Fable 5, the consumer-facing version of the model that had briefly been accessible to general subscribers before the ban. For now, non-approved enterprises, global allies, and everyday consumers remain entirely in the dark regarding when, or if, they will be permitted to access the new architecture. 

Anthropic confirmed that it is actively cooperating with federal officials to design long-term deployment standards. "We're restoring access for these organizations quickly, and we're continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again," the company said in an official statement.