Anthropic received US government approval on June 26 to restore access to its Claude Mythos 5 model for a group of roughly 100 companies and federal agencies, partially ending a two-week standoff over national security concerns linked to the model's advanced cybersecurity capabilities.

Key Highlights

• US Commerce Secretary Howard Lutnick confirmed in a letter to Anthropic that appropriate safeguards are in place to permit trusted partners to access Claude Mythos 5, ending a two-week government-enforced suspension.

• Mythos 5 access was revoked on June 12 under an export control directive citing national security authorities, instructing Anthropic to suspend all access by any foreign national, whether inside or outside the United States.

• Fable 5, Anthropic's public-facing version of the same underlying model, remains restricted, with no timeline confirmed for its restoration.

• Project Glasswing partners, which include firms such as Apple, Microsoft, Nvidia, Cisco, and CrowdStrike, are among the roughly 100 organizations cleared to resume Mythos 5 access.

Anthropic's Claude Mythos 5 is the most capable cybersecurity AI model the company has built, sharing the same underlying architecture as the publicly released Fable 5 but with safety guardrails lifted in specific high-risk areas. The model had been available to a select group of organizations through Anthropic's Project Glasswing program before the US government suspended both models on June 12 under export control authorities, citing concerns that foreign nationals could access frontier AI capabilities with dual-use potential in cyberattack scenarios.

The June 26 clearance, communicated through a letter from Commerce Secretary Howard Lutnick addressed to Anthropic co-founder Tom Brown, who has taken the lead in government negotiations, states that Anthropic's work to address associated risks has yielded significant progress. The approval covers Mythos 5 only. Fable 5, which had been available to the general public, remains offline pending further discussions, and the government is described as moving toward restoring it without a confirmed timeline.

The standoff reflected a broader and ongoing tension between the Trump administration and Anthropic. Earlier this year, the Department of Defense designated Anthropic a supply chain risk following a breakdown in negotiations over whether its models could be used for domestic surveillance and fully autonomous weapons systems. The designation, historically reserved for foreign adversaries, requires defense contractors to certify they will not use Claude models in military work. Anthropic subsequently filed a lawsuit against the administration seeking to reverse the blacklisting, with litigation still active.

On the same day as the Mythos 5 clearance, OpenAI separately announced three new AI models and disclosed it was complying with a government request to limit initial rollout to a small group of trusted partners. The parallel approach across the two leading AI developers signals that US government oversight of frontier model releases has become a structural feature of the current AI deployment environment rather than a one-off intervention.

Project Glasswing partners have collectively uncovered more than 10,000 high or critical security vulnerabilities since Mythos Preview launched in April. For more information on Project Glasswing, visit anthropic.com/glasswing.