President Donald Trump on Friday ordered all U.S. Federal agencies to cease using technology developed by Anthropic, an artificial intelligence firm, escalating a high-stakes dispute that could reshape the future of AI integration within the government. The directive, announced via a post on Trump’s Truth Social platform, instructs agencies to phase out Anthropic products within six months.
The move follows a notification to Anthropic from the government earlier Friday, giving the company until 5:01 p.m. EST to concede to demands from the Pentagon or risk being designated a “supply chain risk” – a label typically reserved for entities considered extensions of foreign adversaries. It remains unclear whether Trump’s announcement supersedes the Pentagon’s threat, according to reports from Reuters and Politico.
At the heart of the conflict lies the Pentagon’s desire for unfettered access to Anthropic’s Claude AI system, which is currently operating on classified networks. Officials want the ability to utilize Claude for “all lawful purposes.” However, Anthropic has maintained two non-negotiable conditions: that Claude not be used in the development of autonomous weapons systems and that it not be employed for mass surveillance of American citizens.
The Pentagon has asserted it has no intention of using AI for either of those purposes and argues it requires the freedom to utilize the technology it is licensing. Anthropic publicly refused to yield on Thursday, with its CEO stating, “Threats will not change our position: we cannot, in good conscience, accede to their request,” as reported by CNBC.
Trump characterized Anthropic’s stance as a “DISASTROUS MISTAKE” and accused the company of attempting to dictate operational procedures to the armed forces in his Truth Social post. Shortly after Trump’s statement, the General Services Administration (GSA) announced Anthropic’s removal from USAi.gov, the federal government’s centralized testing hub for AI tools.
Emil Michael, Under Secretary of Research and Engineering at the Pentagon, indicated to Bloomberg that a deal with Anthropic was “in the final stages” that would “effectively get to where they want to be” before Trump’s announcement. Pentagon spokesperson Sean Parnell, writing on X, stated, “This is a simple and sensible request that will prevent Anthropic from jeopardizing critical military operations and potentially exposing our warfighters to danger. We will not allow ANY company to dictate how we make operational decisions.”
The situation reached a critical point on Tuesday during a meeting at the Pentagon between U.S. Secretary of Defense Pete Hegseth and Anthropic CEO Dario Amodei. While sources described the meeting as cordial, Pentagon officials reportedly threatened to cancel Anthropic’s $200 million contract and imposed the potential supply chain risk designation.
The dispute has drawn attention from other AI leaders, including OpenAI CEO Sam Altman, who reportedly shares Anthropic’s concerns regarding collaboration with the Pentagon. Neither Anthropic nor OpenAI immediately responded to requests for comment from CNN.
Anthropic’s Claude system was the first AI model to operate on the U.S. Military’s classified networks, securing a contract worth up to $200 million last year. Other major AI companies, such as OpenAI, have only secured contracts for unclassified networks. Anthropic’s “acceptable use policy,” outlined in the contract, explicitly prohibits the use of Claude for mass surveillance and autonomous weapons.
Gregory Allen, a senior consultant at the Center for Strategic and International Studies, told Bloomberg Radio that while Claude is popular within the Department of Defense and its usage restrictions haven’t been triggered in conversations he’s had, the Pentagon objects to being constrained by a company’s policies. A Pentagon official told CNN that “you can’t run tactical operations by exception” and that “legality is the Pentagon’s responsibility as the finish user.”
Severing ties with Anthropic could present challenges for the Pentagon, requiring the replacement of any internal systems utilizing Claude. While a Pentagon official stated that Elon Musk’s Grok AI system is “approved for use in a classified environment,” Grok is not considered as advanced as Claude.
The potential financial impact on Anthropic is significant. While losing the $200 million contract wouldn’t be an existential threat to the company, valued at approximately $380 billion, being designated a supply chain risk could be more damaging. This designation would require any company working with the U.S. Military to prove it does not utilize any Anthropic-related products in activities with the Pentagon.
Adam Connor, Vice President of Technology Policy at the Center for American Progress, stated that this could jeopardize a substantial portion of Anthropic’s client base, including companies with existing or potential government contracts. Nvidia CEO Jensen Huang anticipates the Pentagon and Anthropic will reach an agreement, but believes the situation won’t be “the end of the world” even if they don’t, as other AI companies are available.
The Pentagon had also considered invoking the Defense Production Act (DPA), a 1950 law granting the president broad authority to control national industries, to compel Anthropic to cooperate. It remains unclear whether, or how, the Pentagon could force Anthropic to operate with them while simultaneously designating it a supply chain risk. Trump’s statement did not specify whether the DPA would be utilized or if the supply chain risk designation would be imposed.
Connor suggested the Pentagon’s threat serves as a warning to other AI companies seeking to profit from government contracts, signaling they should avoid imposing restrictions on AI usage. Alan Rozenshtein, a law professor at the University of Minnesota, noted that the government’s ultimate goal is to continue using Anthropic’s technology and is employing all available leverage to achieve that.
The Pentagon has stated that if Anthropic does not agree to the terms by 5:01 p.m. EST, the contract will be terminated and the company will be considered a “supply chain risk.” No further public announcements are expected beyond Trump’s statement, and the method for replacing Anthropic’s systems remains undetermined.
