The Trump administration has authorized Anthropic to restore access to Mythos, its most sophisticated AI model, for a carefully chosen group of US organizations. The decision comes after weeks of closed-door negotiations between the White House and the artificial intelligence company. Neither the full list of approved entities nor the specific conditions of the arrangement have been publicly disclosed.
The move signals a shift in how the government is engaging with frontier AI capabilities during a period of heightened scrutiny over national security risks. Critics argue that selective access could create an uneven playing field, while supporters say it allows the US to maintain a competitive edge in AI development. The White House has framed the decision as a measured approach to balancing innovation with security.
Anthropic declined to provide details on the number of organizations now using Mythos or the exact timeline of the restored access. The company previously restricted the model's availability amid regulatory reviews. Wired reported that the arrangement applies to both private-sector firms and government agencies, though no specific names were confirmed by press time.
Industry analysts suggest the limited rollout could accelerate AI adoption in defense and critical infrastructure sectors. However, the exclusion of most commercial and international users raises questions about the administration's broader AI strategy. Some experts warn that gatekeeping advanced models may stifle broader economic benefits.
Critics within the AI safety community have voiced concerns that restricted access concentrates power in too few hands. Without transparent oversight, they argue, the risks of misuse or unintended consequences remain high.