After being disabled, Anthropic's powerful Claude Mythos 5 AI model has been quietly redeployed to just over 100 US cyber defenders and infrastructure providers, bypassing broader public access. The Trump administration allowed Anthropic to make Mythos 5 available to these specific U.S. government agencies and companies. This limited re-enablement marks a new phase in how advanced AI integrates into critical national functions in 2026.
The US government initially restricted access to Anthropic's advanced AI models, but has now selectively eased those restrictions for a specific group of organizations. This tension reveals a direct, hands-on intervention in a private company's product release.
The US appears to be establishing a precedent for a highly controlled, strategic rollout of powerful AI, prioritizing national security applications while potentially stifling wider innovation and public access.
Why the Government is Playing Gatekeeper
US Commerce Secretary Howard Lutnick informed Anthropic that appropriate safeguards are in place for trusted partners to access Mythos, according to WIRED. These safeguards are the mechanism through which the government asserts control over powerful AI, moving beyond simply setting development standards.
The government did not permit a broader rollout of Mythos or comment on the consumer-facing version, Claude Fable 5, as reported by WIRED. This suggests the deliberate creation of a bifurcated AI landscape: a controlled tier for critical applications, and a restricted or delayed public tier. The government's dual concern for both the power and potential risks of advanced AI in critical sectors is evident in the emphasis on 'safeguards' and limited access.
From Restriction to Redeployment: A Quick Turnaround
Anthropic disabled access to its Fable 5 and Mythos 5 models, according to CNBC. This immediate action signaled initial alarm about the models' deployment.
Anthropic acknowledged on X that they had been working with the US government since June 12 to restore access to Mythos 5 and Fable 5, as reported by TechCrunch. This confirmed negotiations were underway. The urgent strategic value the government places on this technology is underscored by the rapid negotiation and conditional re-enablement, even after initial concerns led to its disabling.
The Deal That Brought AI Back Online
Anthropic and the Trump administration reached a deal to bring one of the company's AI models back online. This agreement effectively positioned the state as the ultimate gatekeeper for frontier models, a clear shift from mere regulation to active control over deployment.
By restricting Mythos 5 to "cyber defenders and infrastructure providers," the administration implicitly nationalized access to cutting-edge AI capabilities for strategic national security and critical infrastructure purposes. This 'deal' establishes a new model of public-private partnership in AI, where government oversight dictates access to advanced models for national strategic purposes.
What This Means for Future AI Access
A shift from mere regulation to active control over advanced AI deployment is evident in the government's action of disabling and then selectively restoring access to Mythos 5. This controlled release points to a future where advanced AI models are treated as strategic national assets, with access tightly regulated rather than openly distributed.
This move creates a two-tiered AI ecosystem, where national security interests will dictate who benefits from the most powerful AI advancements. The broader public and companies outside the 'trusted partners' group face denied access to cutting-edge AI, potentially creating controlled scarcity, a dynamic that will shape Anthropic's market access well beyond 2026.
