The White House is working to give federal agencies a clear path to adopt Anthropic’s Mythos AI, the company’s most powerful model to date, according to sources familiar with the matter — a striking about-face from an administration that just months ago helped blacklist Anthropic from the federal government. The draft guidance surrounding Anthropic Mythos AI and its federal deployment is now at the center of one of Washington’s most closely watched technology policy battles. – Restore Anthropic Access.
According to an Axios report published Tuesday, the Trump administration is developing an executive action that would allow government agencies to sidestep an existing supply chain risk designation placed on Anthropic by the Pentagon. That designation, issued after Anthropic refused to strip safety guardrails from its Claude models for military use, effectively made Anthropic a pariah in federal procurement circles — a label previously reserved only for companies linked to foreign adversaries.
Now, sources describe the White House effort as a way to “save face and bring em back in,” as one person familiar with the discussions put it. White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent recently met with Anthropic CEO Dario Amodei in what both sides described as a productive introductory session. President Trump also signaled a warmer stance last week, saying Anthropic was “shaping up” and that a Pentagon deal was possible.
The urgency is driven in no small part by Mythos itself. Anthropic’s newest flagship model has drawn intense interest across the national security community for its reported ability to identify cybersecurity vulnerabilities and devise exploitation strategies — capabilities that the National Security Agency has already moved to leverage, even as the broader feud with the Pentagon plays out in court. The White House is convening companies this week for so-called “table reads” of the proposed Mythos AI deployment guidance, shaping best practices for federal use.
The situation is complicated by an ongoing legal battle. Anthropic filed suit against the Trump administration after the Pentagon issued its supply chain risk designation, and a federal judge in California temporarily blocked non-Pentagon agencies from cutting ties with the company. However, the DC Circuit Court of Appeals later ruled the Defense Department could proceed with severing formal ties while litigation continues, calling it inappropriate to force the military to “prolong its dealings with an unwanted vendor of critical AI services in the middle of a significant ongoing military conflict.”
Meanwhile, competitors have moved to capitalize. OpenAI and Google have both signed agreements allowing the Pentagon to use their AI models under an “all lawful purposes” standard in classified settings — the very terms Anthropic refused. That competitive gap has caused concern among some national security officials who worry that sidelining Anthropic Mythos AI limits America’s edge in a rapidly accelerating global race. – Restore Anthropic Access.
The proposed executive action, if finalized, would stop short of resolving the Pentagon’s core legal dispute with Anthropic. Instead, it would create a pathway for other federal agencies — from the intelligence community to civilian departments — to begin onboarding Mythos under newly defined safeguards. Whether that satisfies agencies clamoring for Mythos access or merely delays a larger reckoning remains to be seen. Anthropic declined to comment on the reported draft guidance.
