Department of Defense certifies AI company Anthropic as ‘effective immediately’ for supply chain risks: NPR

In an unprecedented move, the Trump administration has designated artificial intelligence company Anthropic as a supply chain risk, following through on a threat that could force other government contractors to stop using its AI chatbot Claude.

“We have formally notified Anthropic leadership that the company and its products are considered a supply chain risk, effective immediately,” the Department of Defense said in a statement Thursday.

The decision appeared to close any opportunity for further negotiations with Anthropic, about a week after President Donald Trump and Secretary of Defense Pete Hegseth accused the company of endangering national security.

Trump and Hegseth announced a series of threats last Friday, on the eve of the Iran war, after Anthropic CEO Dario Amodei refused to back down, citing concerns that the company’s products could be used for mass surveillance of Americans or autonomous weapons.

“We do not believe this action is legally sound and see no other option than to challenge it in court,” Amodei said in a statement Thursday.

“This is about the fundamental principle that the military can use technology for any lawful purpose. The military will not allow vendors to enter the chain of command and endanger our nation’s warfighters by restricting the lawful use of critical capabilities,” a Pentagon statement said.

Amodei countered that Anthropic’s narrow exceptions to limits on surveillance and autonomous weapons “relate to high-level areas of use and not to operational decision-making.”

He said there had been “productive conversations” with the Pentagon in recent days about whether to keep Claude in the role or establish a “smooth transition” if no deal is reached. President Trump gave the military six months to phase out Claude, who is already widely integrated into military and national security platforms. Amodei said the priority is to ensure that fighters “are not robbed of critical equipment during large-scale combat operations.”

Some military contractors have already cut ties with tech industry rising star Anthropic, which sells Claude to a variety of companies and government agencies. Lockheed Martin said it would “follow the direction of the President and the Department of the Army” and look to other providers of large-scale language models.

“Lockheed Martin does not rely on a single LLM vendor for any part of its operations and therefore expects the impact to be minimal,” the company said.

It is unclear how the Department of Defense will interpret the scope of the risk designation. Amodei said the notice Anthropic received from the Pentagon on Wednesday indicates it applies only to customers’ use of Claude as a “direct part” of military contracts.

Microsoft said its lawyers are reviewing the rule and that the company “may continue to work with Anthropic on non-defense-related projects.”

Department of Defense decision draws criticism

The Pentagon’s decision to apply rules aimed at addressing supply threats from foreign adversaries was met with widespread criticism. Federal code defines supply chain risk as “the risk that an adversary will sabotage, maliciously introduce unwanted functionality, or otherwise subvert a system for the purpose of destroying, degrading, or espionage.”

Sen. Kirsten Gillibrand, D-N.Y., a member of the Senate Armed Services Committee and the Senate Intelligence Committee, called it a “dangerous misuse of a tool for dealing with adversary-controlled technology.”

“This reckless action is short-sighted, self-destructive and a gift to our adversaries,” he said in a written statement Thursday.

Neil Chilson, a Republican and former chief engineer at the Federal Trade Commission who now heads AI policy at the Abundance Institute, said the decision appears to be a “massive overreach that will harm both the U.S. AI sector and the military’s ability to acquire the best technology for U.S. warplanes.”

Earlier, a group of former defense and national security officials sent a letter to U.S. lawmakers expressing “grave concerns” about the designation.

“The exercise of this power against U.S. domestic companies departs significantly from its intended purpose and sets a dangerous precedent,” said the letter from former officials and policy experts, including former CIA Director Michael Hayden and retired Air Force, Army, and Navy leaders.

It added that such designations are “intended to protect the United States from infiltration by companies beholden to the Chinese government and Moscow rather than from foreign adversaries—American innovators operating transparently under the rule of law. Applying this tool to punish American companies that refuse to lift safeguards against domestic mass surveillance and fully autonomous weapons is a categorical error with implications far beyond this conflict.”

Anthropic sees an increase in consumer downloads

Anthropic lost a major partnership with a defense contractor, but consumer downloads have skyrocketed over the past week thanks to people supporting its moral stance. The company said more than 1 million people signed up for Claude every day this week, making it the top AI app in more than 20 countries, surpassing OpenAI’s ChatGPT and Google’s Gemini in Apple’s app store.

The dispute with the Pentagon also deepens the bitter rivalry between Anthropic and OpenAI that began when former OpenAI leaders, including Amodei, founded Anthropic in 2021.

Hours after the Pentagon punished Anthropic last Friday, OpenAI announced a deal to effectively replace Anthropic with ChatGPT in classified military environments.

OpenAI had called for similar protections for domestic surveillance and fully autonomous weapons, but later said the deal needed to be amended, with CEO Sam Altman saying the company should not have rushed into a deal that “looked opportunistic and sloppy.”

Amodei also expressed regret for his role in that “difficult day for the company,” and said he wanted to “directly apologize” to Anthropic staff Thursday for sending an internal memo to Anthropic staff that attacked OpenAI’s actions and suggested Anthropic would be punished for not giving Trump “dictator-like praise.”