White House Considering Reintroducing Anthropic

According to a report from Axios, the White House is drafting guidance that could allow federal agencies to bypass Anthropic's supply chain risk determination and access new models, including Mythos. Insiders say that an administrative action currently being drafted may contain arrangements related to the government's use of AI.

Anthropic had previously refused to sign an agreement allowing the Pentagon to use Claude for "all lawful purposes," insisting on prohibiting its use for large-scale domestic surveillance or the development of fully autonomous weapons; the Pentagon considers this shows that Anthropic is not a reliable partner and has made a rare supply chain risk determination.

The emergence of Mythos complicates the situation further. The model has demonstrated the capability for automated cyberattacks and could also become an important tool for defenders, putting federal agencies in a dilemma between risk control and capability needs. An informed source said that the White House's efforts are aimed at "saving face and bringing them back on board".

Earlier this month, White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent had met with Anthropic CEO Dario Amodei, with both sides describing the meeting as constructive. This week, the White House is also assembling multiple industry companies to discuss potential administrative actions and best practices for Mythos deployment.

Currently, government agencies, including the Pentagon, can still use Anthropic models during legal disputes, and the National Security Agency is also using Mythos. However, the Pentagon's use of Claude is still based on old service terms that both parties consider overly restrictive, and they do not have access to the model's latest updates.