White House Shifts Toward Deal With Anthropic After Months of Legal Combat
The Justice Department asked a federal judge on Wednesday to pause the government's appeal of a ruling that blocked much of the administration's effort to punish Anthropic. The move signals a significant reversal after months of aggressive action against the AI company, according to six people closely following the dispute.
The Defense Department has not formally dropped its most aggressive attack - Secretary Pete Hegseth's March declaration that Anthropic's ethics policies pose a national security risk. Two federal lawsuits remain unresolved. But the balance of power inside the White House has shifted toward officials who want to end the conflict, sources said.
The Mythos Factor
One major motivation is practical: government agencies want to use Anthropic's Mythos AI model to find and patch software vulnerabilities before adversaries do. The model has drawn widespread concern in the Trump administration, the financial industry, and Europe because of its apparent ability to identify critical security flaws.
Staff from at least two large federal agencies have already reached out to Anthropic about integrating Mythos into their cyber defense efforts, according to a former senior U.S. technology official with direct knowledge of the discussions.
The Office of Management and Budget told government officials in an email last week that agencies may soon be allowed to use a "modified" version of Mythos, according to documents reviewed by POLITICO.
Who's at the Table
A Friday meeting at the White House included chief of staff Susie Wiles, Treasury Secretary Scott Bessent, and National Cyber Director Sean Cairncross. Neither Hegseth nor Defense Undersecretary Emil Michael, who led much of the Pentagon's criticism, attended. The conversation focused on Mythos, not the broader dispute.
Both sides issued optimistic statements after the meeting. President Trump told CNBC on Tuesday that Anthropic executives "came to the White House a few days ago, and we had some very good talks with them." He said the company's leaders are "very smart" and have "high IQs," a sharp departure from his previous characterization of them as "Leftwing nut jobs."
When asked if the administration would make a deal with Anthropic, Trump said: "It's possible."
The Contradiction Problem
The government's position contains an internal contradiction. In court filings, administration lawyers have narrowed the scope of Hegseth's supply chain risk label, saying it only prohibits the Pentagon from working with Anthropic. Yet Trump's February Truth Social post declared that federal agencies must "IMMEDIATELY CEASE all use of Anthropic's technology."
Despite that supposed ban, the Commerce Department's Center for AI Standards and Innovation is already testing Mythos. The Treasury Department has sought access. Other agencies have begun using Anthropic models again.
Dean Ball, a former White House adviser who led the administration's AI Action Plan, said the approach has been "absurd." He added: "We shouldn't let anyone pretend it's not a blaring contradiction, using this thing and also saying that it is the equivalent of a product made by a foreign enemy government."
What Industry Wants
Tech industry representatives say the administration needs a full public reversal to reassure the sector. Paul Lekas, head of global public policy at the Software Information Industry Association, said the government cannot have it both ways - tacitly allowing federal agencies to use Mythos while preserving the national security risk designation on paper.
"The administration is definitely still fighting in court," Ball said. "But I think that clearly there are some people who see the whole thing as a distraction they don't need."
The original conflict began in February when Anthropic CEO Dario Amodei insisted the company would not allow the Pentagon to use Claude AI for mass surveillance or autonomous weapons. Trump and Hegseth accused Anthropic of trying to dictate government policy by imposing limits on what they framed as "lawful use" of AI.
The dispute sent a chill through the AI industry and opened opportunities for Anthropic's rival, OpenAI. A federal judge in northern California temporarily blocked both the supply chain risk label and Trump's directive that agencies stop using Anthropic products, creating legal cover for the current tacit arrangement.
Whether the White House will formally walk back its position remains unclear. Ball said administration officials may be unwilling to completely abandon their animosity toward the company, even as they quietly allow federal use of its most valuable technology.
Your membership also unlocks:
