White House chief of staff Susie Wiles sat down Friday with Anthropic CEO Dario Amodei and Treasury Secretary Scott Bessent in a meeting both sides called productive, a sharp turn from an administration that had publicly vowed to cut ties with the artificial intelligence company over what it labeled "woke" AI.

The sit-down marks the clearest signal yet that the Trump White House is willing to override the Pentagon's hard-line stance against Anthropic, a company the Defense Department tried to blacklist as a national security supply-chain risk. The question now is whether strategic necessity will win out over the political fight that put the company on a federal no-fly list in the first place.

Just the News reported that the White House described the meeting as "both productive and constructive," adding that officials discussed collaboration opportunities and protocols for scaling AI technology. The White House also said it plans to host similar discussions with other leading AI companies.

From blacklist to boardroom

The reversal did not come from nowhere. The Trump administration had previously indicated it would not work with Anthropic, with the president himself posting that the government "will not do business with them again." The Treasury and Defense Departments began stripping Anthropic's AI chatbot, Claude, from their operations. Defense Secretary Pete Hegseth sought to formally designate the company a supply-chain risk, a label that would have effectively locked Anthropic out of the entire federal procurement system.

Anthropic sued the administration. A federal judge then blocked enforcement of Trump's directive against the company, according to AP News.

The administration's core complaint was that Anthropic's built-in safety guardrails, restrictions the company placed on its AI to prevent certain military and surveillance applications, amounted to a national security threat. The argument: a private company should not be able to intervene during military operations by dictating what its technology will and will not do. Anthropic denied the allegations.

That dispute has not been resolved. But something changed the calculus.

Mythos changes the math

The something appears to be Anthropic's newest AI model, called Mythos. Fox News reported that the model is described as highly advanced and potentially important for cyber defense, prompting renewed interest from the administration despite prior political and national security objections. Trump had ordered a government-wide halt on Anthropic technology after the Pentagon clash, but the White House is now reconsidering that stance.

The New York Post reported that the administration had been discussing providing government agencies advance access to Mythos. Senior officials, including Vice President JD Vance, Treasury Secretary Bessent, and Defense Secretary Hegseth, convened to prepare for the possibility, the Post noted. Some intelligence and cybersecurity agencies were already testing the technology.

The strategic logic is not subtle. A source close to the talks told Axios, in a quote carried by multiple outlets:

"It would be grossly irresponsible for the U.S. government to deprive itself of the technological leaps that the new model presents. It would be a gift to China."

That framing, lose Anthropic, lose ground to Beijing, appears to have concentrated minds inside the West Wing. The administration has consistently treated the AI race with China as a top-tier national security priority, and sidelining one of America's most capable AI firms over a policy disagreement sits uneasily with that posture.

The tension mirrors other moments where the White House has had to weigh political commitments against operational realities. The administration has faced similar dynamics in its interactions with the Pentagon over Iran policy, where competing institutional priorities forced difficult choices at the top.

What both sides said

The White House statement, provided to The Hill, laid out the official version of the meeting's scope:

"We discussed opportunities for collaboration, as well as shared approaches and protocols to address the challenges associated with scaling this technology. The conversation also explored the balance between advancing innovation and ensuring safety. We look forward to continuing this dialogue and will host similar discussions with other leading AI companies."

An Anthropic spokesperson offered a parallel account. Fox News Digital carried the company's statement describing the meeting as a "productive discussion on how Anthropic and the U.S. government can work together on key shared priorities such as cybersecurity, America's lead in the AI race, and AI safety."

Neither statement addressed the lawsuit, the blacklist, or the underlying dispute over AI safeguards in military applications. Both read like the kind of diplomatic boilerplate that follows a meeting where the parties have agreed to stop fighting, at least publicly, without yet agreeing on terms.

A White House official added, as Breitbart reported, that the administration is "engaging with advanced AI labs about their models and the security of software," though any federal use of Anthropic's technology would require a technical evaluation period.

The safeguards question remains

The central tension has not gone away. The administration argued that Anthropic's refusal to remove safeguards on autonomous weapons and mass surveillance applications made the company unsuitable for defense work. Anthropic pushed back, insisting those guardrails were responsible design, not political sabotage.

That disagreement matters beyond Anthropic. It touches a question every AI company building tools for the federal government will eventually face: who decides what an AI system will and will not do, the company that built it, or the government that bought it?

The administration's position has a straightforward logic. If the military deploys an AI tool in a combat or intelligence scenario, the tool needs to work. A company that builds in restrictions on how the government can use its product is, in the administration's view, inserting itself into the chain of command. The national security implications of that arrangement are real.

Anthropic's position also has a logic, even if this publication finds it less persuasive. The company argues that unrestricted AI in military settings creates risks the technology is not yet mature enough to handle safely. Whether that concern is genuine engineering caution or ideological gatekeeping dressed up as safety is exactly what the dispute is about.

The White House has navigated similar tensions between institutional resistance and executive priorities on other fronts, including moments when establishment pressure campaigns tested the administration's resolve on personnel and policy.

What comes next

The meeting between Wiles, Bessent, and Amodei does not settle the dispute. It opens a channel. The White House said it would host similar discussions with other AI companies, suggesting the administration wants to be seen as engaging the entire industry, not cutting a special deal with a firm it recently tried to exile.

Whether Anthropic's Mythos model ends up in federal hands depends on the technical evaluation the White House official referenced. It also depends on whether Anthropic is willing to modify or remove the safeguards that triggered the fight in the first place. The company has given no public indication it will.

Newsmax reported that the administration was weighing the value of Anthropic's technology against cybersecurity concerns, with some intelligence agencies already testing it. That suggests the practical question, is this technology too good to pass up?, may already be answering itself inside the national security apparatus, even as the political and legal questions remain unresolved.

The broader AI competition with China adds urgency. The administration has treated technological supremacy as a core national interest, and the logic of that position creates pressure to work with every capable American AI firm, even the ones that annoyed you. Decisions about countering China's strategic ambitions have already forced the administration to make pragmatic moves in other arenas.

Several questions remain unanswered. What specific terms, if any, were discussed for Anthropic's reentry into federal work? Will the company adjust its safeguards as a condition of collaboration? What happens to the lawsuit? And does the Pentagon, which started this fight, get a veto over whatever the White House negotiates?

None of those answers came out of Friday's meeting. What came out was a handshake and a press release. That is more than Anthropic had a week ago.

When national security and political grudges collide, the smart bet is usually on national security. The administration appears to be making that bet, and if Anthropic's technology is as advanced as both sides seem to believe, it is the right one. But the company that built restrictions into a tool the military needs should not expect the terms to be generous.