Anthropic and multiple parts of the U.S. government have begun to repair a strained relationship that ruptured earlier this year, according to people familiar with the interactions. The company, which develops advanced AI systems, is preparing for a stock-market debut that some estimates suggest could value it at $1 trillion. The timing has coincided with efforts on both sides to reduce public tensions.
The disagreement began after Anthropic refused requests from the U.S. military to permit use of its AI models for domestic surveillance and for fully autonomous weapons systems. In response, federal officials moved to impose a national security blacklist on the company that is scheduled to take effect later in the year. In March, the Defense Department labeled Anthropic a "supply-chain risk" - a designation typically used for firms linked to adversarial nations - making it the first U.S. company to receive that classification. That label prevents tens of thousands of government contractors from using Anthropic's AI on defense-related work.
Despite the formal designation, interactions have resumed at senior levels. Anthropic CEO Dario Amodei visited the White House in mid-April, marking the first such engagement since the dispute emerged. The White House also extended an invitation for Amodei to attend a planned May 21 signing ceremony for an executive order on artificial intelligence, according to two people familiar with the matter. That event was ultimately canceled - sources say the cancellation was linked to the president's objections to particular elements of the draft order - but the executive order was signed on Tuesday.
In a statement posted on X, Anthropic said it looked forward to "collaborating" with the administration on implementing the order. A White House spokesperson did not respond to a request for comment about the relationship between the administration and the company, and Anthropic's press team declined to comment.
Cybersecurity briefings and technical discussions
Beyond public gestures, Anthropic has held substantive discussions with National Cyber Director Sean Cairncross about Mythos, the company's most advanced AI system, and the risks it could pose to critical infrastructure by amplifying cyberattacks, a person familiar with the discussions said. Anthropic itself has warned that Mythos could "supercharge" such attacks. In this context, critical infrastructure refers to key economic sectors - such as banks, emergency services and hospitals - that receive special protection against disruptive attacks.
The company also met with U.S. Treasury Secretary Scott Bessent this spring to discuss Mythos and potential presidential actions on AI, according to a U.S. official. That engagement was reported to have informed elements of the administration's executive order issued on June 2, including a request that leading model developers provide their most advanced systems for cybersecurity testing.
Legal battle and continued exclusions
Legal friction remains. Anthropic is challenging the Defense Department's supply-chain risk designation in court, and the department is vigorously defending the label, according to Franklin Turner, an attorney who specializes in government contracts. Turner cautioned that until the dispute with the department is resolved, broader damage to Anthropic's business is unlikely to abate. "Anytime the government signals that it's washing its hands of a company, that's a major problem for that company," he said.
Practical exclusions persist as well. Anthropic personnel were not part of an April 27 Army-run AI simulation focused on cyberattacks, according to Army spokesperson Sean Minton. That simulation included cyber executives from leading AI developers, among them Google and OpenAI, according to an Army statement. Both sides were still contesting the supply-chain risk designation as of Thursday, when they filed briefs with the court.
Market and corporate considerations
Anthropic's efforts to mend ties come as it prepares for a high-profile initial public offering. Executives and outside corporate leaders have moved to solidify relationships with the White House in part to avoid running afoul of presidential displeasure. PitchBook senior research analyst Harrison Rolfes, who covers AI firms, said that stronger ties to the administration could help restore investor confidence in the short term and described the public dispute as a "near-term bruise."
For now, the picture is mixed: White House-level engagement and technical briefings point to a softening of tensions, while the statutory designation and the ongoing litigation with the Defense Department underscore unresolved legal and operational obstacles that could continue to affect Anthropic's access to government contracts and partnerships.