The Trump administration is quietly recalibrating its approach to Anthropic, the artificial intelligence firm it blacklisted from military and government work earlier this year, as the company's new Mythos model proves too valuable for federal agencies to ignore.
Mythos, Anthropic's most advanced AI system to date, can identify decades-old security vulnerabilities in major web browsers and software. The model has drawn interest from the National Security Agency, the Treasury Department, and other federal bodies, creating an opening for the company to mend its fractured relationship with the White House. The shift is notable given that just two months ago, President Trump publicly labeled Anthropic an "out of control radical left" company, casting doubt on its future in a Republican-led Washington.
But in recent weeks, the tone has changed. Trump told CNBC on Tuesday that while Anthropic "tends to be on the left," he added, "we get along with them." He praised the firm's talent, saying, "They're very smart, and I think they can be of great use. I like smart people, I like high-IQ people and they definitely have high IQs." The president also suggested a deal to use Anthropic in the Defense Department is "possible."
Behind the scenes, Anthropic CEO Dario Amodei has held high-level talks at the White House with chief of staff Susie Wiles and reportedly Treasury Secretary Scott Bessent, focusing on Mythos and broader cybersecurity concerns. The White House described the discussions as "very good talks," with Trump noting the company is "shaping up."
The rapprochement comes even as Anthropic fights the government in two separate courts. The company sued the Pentagon last month over its designation of Anthropic as a "supply chain risk"—a label typically applied to firms linked to foreign adversaries—after negotiations over safety guardrails collapsed. That designation prohibits the Defense Department from using Anthropic's products. A separate legal challenge targets Trump's social media directive to civilian agencies to stop using the firm's technology. A California judge granted a preliminary injunction against both actions, but a Washington, D.C., appeals court panel rejected Anthropic's request for an emergency stay.
Despite the ongoing litigation, Mythos has become a bargaining chip. The company released the model to a limited group of technology firms, cybersecurity companies, and Wall Street banks, prompting speculation about whether the government would gain access. Jessica Tillipman, associate dean for government procurement law studies at George Washington University Law School, said, "It really does feel like this is the government cutting off its nose to spite its face."
Federal agencies have since moved to secure access for defensive work. An Anthropic official confirmed the company has briefed teams at the National Institute of Standards and Technology and the Cybersecurity and Infrastructure Security Agency. Axios reported that the National Security Agency, which falls under the Pentagon, is already using Mythos despite the ban, while the Treasury Department has formally requested access. Bessent and Federal Reserve Chair Jerome Powell also met with Wall Street executives on the issue, and Bessent joined Vice President Vance for a call with Amodei and other AI leaders to discuss cybersecurity risks associated with advanced models.
Dean Ball, a co-author of Trump's White House AI Action Plan who supported Anthropic in its clash with the Pentagon, said the Mythos release underscores the administration's earlier missteps. "I think the gods are trying to tell us something about the correct way that we are dealing with the government," Ball said. "You can't say that it looked particularly wise."
Technology experts argue that the administration's ideological objections may be unsustainable given the concentration of frontier AI capabilities. Anthropic is one of only a handful of companies producing the country's most advanced models. "The government is on some level losing some of its leverage because there are only a handful of frontier models that they can tap into. It's a highly concentrated market," Tillipman said, adding that the administration appeared to be "grasping for whatever" justification to limit access.
