Anthropic restores Mythos 5 access for select US organizations

The United States government has partially lifted the restrictions imposed on Anthropic’s advanced AI model, Claude Mythos 5, following two weeks of negotiations between the company and federal officials. On June 26, 2026, U.S. Commerce Secretary Howard Lutnick issued a letter to Anthropic’s chief compute officer, Tom Brown, confirming that the government would permit the redeployment of the model to a select group of trusted partners.

This decision ends a period of total blackout for the firm’s most capable cybersecurity-oriented systems. The initial directive, issued by the Department of Commerce on June 12, had mandated that Anthropic cease all access to both Mythos 5 and its consumer-facing counterpart, Fable 5. That order, anchored in export control authorities, prohibited access by foreign nationals — a group that included some of Anthropic’s own staff as well as employees at partner organizations. Because the company was unable to restrict access to those specific individuals quickly enough, the models were effectively pulled offline entirely.

Media additions

U.S. government gives Anthropic green light for limited re-release of Mythos 5

The restoration of Mythos 5 applies to more than 100 organizations that operate and defend critical infrastructure. Under the new arrangement, these entities — along with Anthropic, may now grant access to their foreign national employees. While this represents a significant thaw in relations, the directive does not apply to Fable 5, which remains offline with no confirmed timeline for public release.

Safety Concerns and Regulatory Scrutiny

The standoff began amid growing federal anxiety regarding the models' capabilities. Administration officials had expressed concerns that the systems could be “jailbroken” to bypass safety filters, potentially allowing users to weaponize the technology. According to reporting, these concerns were compounded by reports that the models were granted access to a South Korean telecommunications firm believed to have ties to China. Furthermore, the National Security Agency and other departments raised alarms about the possibility of the model identifying vulnerabilities across classified computer systems.

In his June 26 letter, Secretary Lutnick stated that the government now has confidence in the safeguards implemented by Anthropic.

"Anthropic has worked with the U.S. government to address risks associated with the Covered Models. These efforts have yielded significant progress,"

Howard Lutnick, Commerce Secretary, via Wired

Despite this progress, the government’s approach remains cautious. The June 26 letter explicitly notes that all other requirements outlined in the original June 12 directive remain in effect, and the Commerce Department reserves the right to adjust license requirements should circumstances change.

Market and Industry Impact

The two-week suspension has occurred during a sensitive period for Anthropic, which filed a confidential S-1 with the SEC on June 1 in preparation for a proposed initial public offering. The blackout allowed competitors to gain ground. For example, the Chinese firm Zhipu launched its GLM 5.2 model one day after the U.S. Export ban went into effect.

The regulatory environment for these companies is shifting toward an incremental, government-approved model. OpenAI, which announced the staggered release of its GPT-5.6 models on June 26, is now operating under a similar framework where the federal government vets and clears trusted partners on a case-by-case basis. In its release announcement, OpenAI stated, We don’t believe this kind of government access process should become the long-term default.

What to Watch Next

• Fable 5 Status: While Mythos 5 is returning to select partners, the public-facing Fable 5 model remains unavailable. Negotiations between Anthropic and the White House regarding this version are expected to continue through the weekend.
• Long-term Frameworks: Both Anthropic and OpenAI are pushing to work with the administration to move away from the current ad-hoc, case-by-case approval process toward a more repeatable policy framework for future model releases.
• Cybersecurity Protocols: Under the latest agreement, Anthropic has committed to continue working with the U.S. Government on standardized protocols for safety testing and model deployment.
• Corporate Development: With the S-1 filing public, the company’s ability to stabilize its product roadmap and resolve the Fable 5 situation remains a critical factor for its ongoing growth trajectory.

As of June 27, 2026, Anthropic has stated it is working to provision the approved set of providers and restore their access to Mythos 5 as quickly as possible. The company maintains that it continues to work with the government to expand access to Mythos 5 and reinstate Fable 5 for general use.