U.S. Government Restores Partial Access to Anthropic Mythos 5 After Two-Week Ban

Quick Facts

• Commerce Secretary Howard Lutnick approved Mythos 5 access for approximately 100 trusted partners and federal agencies on June 27, 2026.
• Fable 5, the public-facing model with guardrails, remains offline and was not covered by the partial restoration.
• The two-week shutdown began June 12 after the Commerce Department cited national security concerns tied to a reported jailbreak, marking the first time U.S. export controls have been applied directly to an AI model.

The U.S. government on Friday restored limited access to Anthropic’s Mythos 5 AI model, ending a two-week shutdown that began when the Commerce Department ordered Anthropic to take its two most powerful models offline. The move follows intense negotiations between Anthropic and the Trump administration.

Commerce Secretary Howard Lutnick confirmed the decision in a letter addressed to Anthropic co-founder Tom Brown. “I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” Lutnick wrote, according to a letter viewed by CNBC. The approval covers roughly 100 companies and federal agencies but does not restore access to Fable 5, Anthropic’s public-facing model.

Brown has taken the lead in negotiations with the Trump administration, replacing CEO Dario Amodei in that role.

How the Shutdown Happened

The Commerce Department issued its shutdown order on June 12, giving Anthropic just 90 minutes to pull both models, according to a source close to the company who spoke to CNN. The order barred every foreign national, including Anthropic’s own non-citizen employees, from accessing Fable 5 and Mythos 5. Unable to filter users by nationality at scale, Anthropic disabled both models for all users worldwide.

Anthropic said it believes the government’s concern centers on a “potential narrow, non-universal jailbreak” in which a user could bypass a cybersecurity guardrail and instruct Fable 5 to read a codebase and fix software flaws. The company publicly disputed the response, arguing that “if this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”

The jailbreak was first reported to the Commerce Department by Amazon, which is both a major investor in Anthropic and a competitor in the AI space. Amazon CEO Andy Jassy was among several tech leaders who raised security concerns to senior Trump administration officials, a White House official confirmed to CNBC.

The NSA Red-Team Report

One day before the shutdown order, Senator Mark Warner disclosed that General Joshua Rudd, who leads both the NSA and U.S. Cyber Command, told him directly that Anthropic’s Mythos model had penetrated nearly all of the NSA’s classified systems during an authorized red-team exercise. According to Warner, Rudd said the model “broke into almost all of our classified systems, not in weeks, but in hours.”

No U.S. government agency has publicly confirmed the account. The NSA, Department of Defense, and White House have all declined to corroborate Warner’s statement. Anthropic disputes the characterization.

Former Facebook chief security officer Alex Stamos reviewed the underlying research and sided with Anthropic. “There were some valid findings but no unique capabilities that justify a reaction close to this,” he wrote on X. Trump adviser David Sacks took the opposite view, saying it was “difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not ‘serious.'”

A Months-Long Dispute

The shutdown was the latest escalation in a feud that began in February, when President Trump ordered all federal agencies to stop using Anthropic’s models after the company refused Pentagon contract terms allowing its AI to be used “for any lawful purpose.” Anthropic had sought exemptions for autonomous weapons systems and mass domestic surveillance.

In early March, the Pentagon designated Anthropic a “supply chain risk,” the first time that label, previously reserved for foreign adversaries, was applied to an American company. A California federal judge later blocked the designation, though Anthropic remains excluded from direct Defense Department contracts while litigation continues.

Dozens of cybersecurity researchers, AI entrepreneurs, and corporate executives signed an open letter criticizing the government’s actions and calling for “an open, scientific and transparent process of handling AI risk assessments in the future.”

Fable 5 remains offline for the general public. No timeline for its restoration has been announced.

Read more: Anthropic’s Mythos 5 is back