US Department of Defense Officially Stops Using Anthropic's AI Tools

Image source, Reuters

An official has told the BBC that the US Department of Defense has stopped using Anthropic's artificial intelligence (AI) tools. Several months ago, the Pentagon designated the company as a supply chain risk on national security grounds.

In a statement on Monday (October 5), the defense official said the Pentagon "has ceased using Anthropic's products."

Defense Secretary Hegseth designated Anthropic as a supply chain risk in February of this year and announced that the Pentagon would stop using its products by late August. It is not yet clear why the delay occurred.

Multiple people familiar with the matter told the BBC that until last week, the department was still using Anthropic's Claude model in research, analysis, intelligence gathering and military operations against Iran.

Trump rules out US-China cooperation on AI development, September 30, 2026; Anthropic co-founder calls for mandatory AI "emergency shutdown" mechanism, September 15, 2026; Trump-Xi summit: the issues left unsaid may be more important, September 29, 2026

End of content

People familiar with the matter said Claude was seen as a key tool for sifting through vast amounts of information. They include former US defense officials and contractors who have worked closely with the Pentagon in the field of artificial intelligence.

The BBC asked Anthropic for confirmation regarding the Pentagon's statement, and a company spokesperson declined to comment.

The details of the Defense Department's full discontinuation of Anthropic's tools had not previously been reported.

Since 2024, Anthropic's products have been used by the US government and military, and it is also the first advanced AI company whose tools have been deployed to classified government agencies.

Chinese-American real estate agent arrested on suspicion of surveilling Lai Ching-te's son, accused of acting as a Chinese agent; Taipei gay bar searched, police officers' discriminatory remarks spark controversy; Taipei mayoral race: Chiang Wan-an's 2028 outpost battle, can Shen Po-yang break through with the "ocean current"?; "Not again": US Marine arrested on suspicion of murdering a woman, reigniting protest wave in Okinawa, Japan

End of Trending

But earlier this year, the Pentagon pressured the company to remove Claude's safety guardrails so that the military could use its AI products without restrictions.

Anthropic refused, citing concerns about mass surveillance and autonomous weapons. Hegseth then designated the company as a "national security supply chain risk" — a legal measure usually applied only to companies headquartered in countries that pose a threat to the United States.

Anthropic called the move "unprecedented and unlawful" and has sued the Trump (Donald Trump) administration in an attempt to overturn the designation.

Image source, EPA

As the legal proceedings move forward, the Pentagon has signed contracts with other major tech companies such as Google, xAI and OpenAI.

According to two people familiar with the Defense Department's AI use, OpenAI's tools have been more widely adopted in some military departments in recent months.

The Pentagon's statement on Monday was in response to BBC questions about its use of Anthropic's products. Previously, multiple interviewees said Claude was still being widely used.

Multiple sources said Claude is embedded in a larger data platform used by the Pentagon — the Maven Smart System.

Maven is operated by Palantir and is the Pentagon's main platform for integrating intelligence and other data.

Lauren Kahn, a senior research analyst at Georgetown University's Center for Security and Emerging Technology, said that because Claude is deeply embedded in Maven, it is not easy to quickly remove it from the system.

"The fact that the Pentagon is only now fully removing Claude from the entire system shows that these kinds of systems are not plug-and-play," she said.

"Once they are integrated, pulling them out again is painful," added Kahn, a former US defense official.

The BBC interviewed defense contractors specializing in AI who work closely with the Pentagon, as well as former US defense officials and others familiar with the Pentagon's use of artificial intelligence. Several of them requested anonymity because they were not authorized to speak to the media.

A former senior US Defense Department official said that when Hegseth initially ordered the phasing out of Anthropic across government departments, "there were concerns about whether the transition from Anthropic to OpenAI" and other tools would go smoothly.

On February 27, Hegseth set a six-month deadline for phasing out Anthropic, and the next day the United States and Israel launched attacks on Iran.

A person familiar with the matter said that "throughout the entire process" of the dispute with the company, Pentagon employees continued to use Anthropic's Claude tools, including its powerful Mythos model.

Multiple sources said Claude was specifically used for intelligence gathering and analysis, including through Palantir's Maven system. Palantir is a leading defense technology company founded by Peter Thiel, a billionaire ally of President Donald Trump.

Sources familiar with how the technology is used said analysts and other defense officials rely on Maven to obtain and organize visual data, including satellite imagery and drone footage.

This data is then fed into Claude and other large language models for tasks such as identifying potential military targets.

A source who received a Palantir demonstration of Maven earlier this year said officials frequently use the technology when writing one-page briefs sent up the chain of command.

Some US allies also use Maven to share intelligence with the Trump administration.

The clash between Anthropic and the Pentagon comes as the company's relationship with the White House appears to be warming. Last week, Trump met twice at the White House with Anthropic chief Amodei.

"How to deal with these risks is still under discussion," Amodei told reporters at the White House, with Trump standing beside him, "but if we work with the president, we can win safely."

Trump then praised Amodei, calling him "amazing" and "terrific."

This article was originally written in English. We used artificial intelligence to assist with translation, and it was reviewed by a BBC journalist before publication. Learn more about how we use artificial intelligence.

The Pentagon's suspension of Anthropic does not mean the US military is withdrawing from the AI military race. The core of the dispute is: between the military and tech companies, who has the right to decide how the technology is used. Anthropic refused to give up restrictions on mass domestic surveillance and fully autonomous weapons, while the US Department of Defense demanded acceptance of "any lawful use."

This is also not a confrontation between an AI company opposed to military use and the military. Anthropic has stated it supports the use of AI for lawful foreign intelligence and counterintelligence missions. Sources told the BBC in February that Anthropic's large model Claude had been used in the US operation to capture Venezuelan President Nicolás Maduro.

By contrast, AI's role in the Iran conflict is clearer. Louis Mosley, Palantir's head of UK and Europe, said in an exclusive BBC interview in April that its Maven Smart System "played a role" in combat management.

According to reports, the Maven Smart System gathers intelligence, satellite imagery and drone footage, analyzes them and proposes strike targets, and can also recommend what level of force to use based on available troops and equipment. Claude is one of the AI large models underpinning the system.

Although Palantir stresses that final decisions are still made by humans. But experts interviewed by the BBC said that once speed and strike scale become priorities, military personnel may not have enough time to verify targets, and may even become overly reliant on the software's judgments. The controversy therefore goes beyond "whether a human presses the button" to whether operators can truly understand and review AI recommendations.

The broader backdrop to this controversy remains competition with China. The US National Security Commission on Artificial Intelligence once opposed a blanket ban on AI-driven autonomous weapons and warned that if China deploys AI-driven swarm drone systems first, the US could lose its military technological advantage. Critics argue that such claims could fuel an irresponsible arms race.

In February this year, when US Deputy Defense Secretary Emil Michael was asked why the Pentagon would not accept Anthropic's contract restrictions, he said: "We do have to prepare for what China is doing."

Image source, Getty Images

Limited reporting also sheds light on China's efforts. Reuters reported in 2024, citing research papers, that a team including researchers from an institution under the PLA Academy of Military Sciences used Meta's publicly released early Llama model to develop a tool called ChatBIT, studying its use in intelligence processing and combat decision support.

Another path is unmanned equipment. At the September 3 military parade in 2025, China displayed various unmanned systems, including unmanned underwater vehicles, drones and "robot wolves."

Beyond competition, China and the US have also drawn a common bottom line. In November 2024, Xi Jinping and then-US President Joe Biden reached a consensus: the decision to use nuclear weapons must remain in human hands and cannot be handed over to AI. But this is not a ban on all military AI, nor does it clarify the boundaries for use in conventional weapons, intelligence analysis and target identification.

The Anthropic incident highlights precisely the blurry boundaries of AI militarization: governments, citing the need to counter rivals, demand greater usage rights; companies try to hold the line on restrictions. Even if final decisions are still made by humans, how to prevent people from becoming mere "approvers" of AI recommendations in high-speed decision-making remains an unavoidable question for the military and technology suppliers.

Explore more

US Department of Defense Officially Stops Using Anthropic's AI Tools

Why read this

Topics

Tags

Body · AI translation