yoga-mahendra
The US Department of Defence has stopped using artificial intelligence products made by Anthropic, a department official has confirmed to the BBC, bringing to an end a relationship that had made the company the first advanced AI developer to have its tools deployed across classified government work.
The confirmation came only after the BBC pressed the Pentagon for comment, following interviews with multiple people who said Anthropic's Claude model was still in active use as recently as last week. A spokesman for Anthropic declined to comment on the department's statement. The development had not previously been reported.
The presence of Anthropic's technology inside the Pentagon was not a recent arrangement. The company's tools had been used by US government agencies and the military since 2024, and it held the distinction of being the first cutting-edge AI firm to have its models running inside government departments handling classified material.
That position began to unravel earlier this year, when the Pentagon asked Anthropic to strip safety guardrails from Claude and grant the military unrestricted access to its systems. Anthropic declined, pointing to concerns about mass surveillance and the use of autonomous weapons.
The refusal triggered a sharp response from Defence Secretary Pete Hegseth, who in February designated the company a supply-chain risk on national security grounds — a legal classification typically reserved for firms based in countries hostile to the United States. Hegseth also publicly criticised Anthropic's chief executive, Dario Amodei.
Anthropic called the designation "unprecedented and unlawful" and responded by suing the Trump administration in an attempt to have it overturned. The company has also been involved in separate legal action, with a judge later ruling that the administration had illegally retaliated against it. The legal fight has continued in the courts while the Pentagon's operational ties to the company quietly persisted.
On 27 February, Hegseth set a six-month deadline to phase Anthropic out of government use — one day before the United States and Israel launched strikes against Iran. According to multiple people familiar with the matter, the department continued to rely on Claude throughout the entire period of the controversy, including during military operations against Iran.
Those sources, who include former US defence officials and contractors who have worked closely with the Pentagon on AI, said Claude was viewed as a critical tool for processing enormous volumes of information. It was used in research, analysis and intelligence gathering, and some of the people said Anthropic's powerful Mythos models remained in use as well.
Much of that activity flowed through a platform called Maven Smart System, the Pentagon's primary system for organising intelligence and other data. Maven is operated by Palantir, the defence technology company founded by Peter Thiel, a billionaire ally of President Donald Trump. Sources said Claude was embedded within Maven rather than being used as a standalone tool.
According to people familiar with how the technology was applied, analysts and defence officials used Maven to access and organise visual material, including satellite imagery and drone footage. That data was then fed into Claude and other large language models for tasks such as identifying potential military targets. Officials also frequently drew on the technology when drafting one-page briefs that moved up the chain of command, according to a source who was given a demonstration of Maven by Palantir earlier this year.
Lauren Kahn, a senior research analyst at Georgetown's Center for Security and Emerging Technology, said Claude's deep integration into Maven would have made a rapid removal difficult. "The fact that it took the Pentagon until now to remove all use of Claude from across the system is indicative that these things are not just plug and play," she said. "Once they become integrated it can be painful to remove them," added Kahn, a former US defence official.
The phase-out has unfolded against a backdrop of the Pentagon broadening its relationships with other major technology companies. It has signed contracts with Google, xAI and OpenAI. OpenAI's tools have been adopted more widely across some military departments in recent months, according to two people familiar with AI use at the defence department.
A former senior US defence official said that when Hegseth first ordered Anthropic to be phased out, "there was concern with moving beyond Anthropic to OpenAI" and other available tools — suggesting that replacing the company's capabilities was not a straightforward substitution.
Anthropic has separately drawn attention for its own safety work. Last September, the company said it had blocked a possible attempt to use AI to make biological weapons, and Amodei publicly called for the pace of AI development to slow down. Trump recently met with Amodei at the White House, an encounter that took place as the legal dispute between the company and his administration continued.
The people who spoke to the BBC did so largely on condition of anonymity, citing a lack of authorisation to speak with the media. Their accounts, taken together with the Pentagon's statement, describe a gap of several months between the political decision to cut ties with Anthropic and the practical reality of removing its technology from systems that military officials had come to depend on.
Comments (0)