lukman-hakim
The US Department of Defence has confirmed it is no longer using artificial intelligence tools developed by Anthropic, according to a statement provided to the BBC. The decision brings to a close a months-long standoff between the Pentagon and one of the country's leading AI developers, a dispute that has raised difficult questions about the role of commercial technology in military operations and the limits of corporate responsibility in the age of autonomous warfare.
A department official said the Pentagon "has ceased the use of Anthropic products," a statement that followed repeated questions from the BBC about the company's continued presence inside defence systems. The confirmation came after multiple people familiar with the matter said Anthropic's Claude model was still being used as recently as last week across research, analysis and intelligence gathering, as well as in military operations connected to Iran.
Those sources, including former US defence officials and contractors who have worked closely with the Pentagon on AI, described Claude as a critical tool for processing vast quantities of information. They said the model had become embedded in workflows that defence analysts and commanders relied upon daily. A spokesman for Anthropic declined to comment when asked about the Pentagon's statement, and the details of the department's full withdrawal from Anthropic's tools had not previously been reported.
The relationship between Anthropic and the US government dates back to 2024, when the company became the first advanced AI developer to have its tools deployed in government agencies carrying out classified work. That arrangement placed Anthropic at the centre of the American defence establishment's accelerating adoption of machine learning, a shift that has transformed how intelligence is sorted, analysed and acted upon.
The break came earlier this year, when the Pentagon pressed Anthropic to strip safety guardrails from Claude and grant the military unrestricted access to its AI tools. Anthropic refused, citing concerns about mass surveillance and the development of autonomous weapons. In response, Defence Secretary Pete Hegseth designated the company a "national security supply-chain risk" — a classification typically reserved for firms based in countries considered hostile to the United States.
Anthropic described the move as "unprecedented and unlawful" and has since sued the Trump administration in an effort to have the designation overturned. The legal fight has played out in the courts even as the Pentagon moved to sign contracts with other major technology companies, including Google, xAI and OpenAI. According to two people familiar with AI use inside the defence department, OpenAI's tools have been more widely adopted in some military departments in recent months.
Hegseth set a six-month deadline to phase out Anthropic on 27 February, one day before the US and Israel launched attacks on Iran. Despite that order, Pentagon employees continued to use Claude, including its powerful Mythos models, "throughout the whole period of the controversy," one person said. The persistence of that use underscores how deeply the technology had been woven into defence operations.
Much of that integration ran through Maven Smart System, a platform operated by Palantir and described as the Pentagon's primary tool for organising intelligence and other data. Several sources said Claude was embedded within Maven, which analysts and defence officials use to access and organise visual data, 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 frequently relied on the technology when compiling one-page briefs passed up the chain of command, according to a source who was given a demonstration of Maven by Palantir earlier this year. Palantir, a leading defence technology firm, was founded by Peter Thiel, a billionaire ally of President Donald Trump, a connection that has added a political dimension to the debate over how military AI contracts are awarded and overseen.
Lauren Kahn, a senior research analyst at Georgetown's Center for Security and Emerging Technology, said the depth of Claude's integration into Maven meant removing it quickly would have been 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 BBC spoke with defence contractors specialising in AI who have worked closely with the Pentagon, as well as former US defence officials and others familiar with the department's use of artificial intelligence. Several spoke on condition of anonymity because they were not authorised to speak to the media. One former senior US defence official said that when Hegseth initially ordered Anthropic to be phased out, "there was concern with moving beyond Anthropic to OpenAI" and other tools.
The episode highlights a broader tension facing governments as they race to adopt AI for national security purposes. Military planners want access to the most capable models available, but many of the companies building them have established ethical guidelines that restrict how their technology can be used. When those guidelines collide with military demands, the result can be contractual disputes, regulatory action and courtroom battles that unfold while operations continue.
For Anthropic, the confrontation has already carried significant consequences. The supply-chain risk designation, normally applied to firms from adversary nations, effectively barred the company from federal contracts and signalled that its refusal to loosen safety controls carried a cost. The company's lawsuit argues that the designation was retaliatory and exceeded the government's legal authority, a claim that a judge has already partly validated in a separate ruling concerning the administration's conduct.
Trump recently met with Anthropic boss Dario Amodei at the White House, a encounter that suggests the relationship between the company and the administration remains fluid despite the legal hostilities. Anthropic has also been active on other fronts, blocking a possible attempt to use AI to make biological weapons and publicly calling for the pace of AI development to slow down.
The Pentagon's confirmation marks the end of one chapter in a dispute that is likely to shape how the US military procures and deploys artificial intelligence for years to come. It also leaves open the question of whether the guardrails Anthropic fought to preserve will survive in the systems that replace Claude, or whether the pressure to field the most powerful tools available will erode the safety standards the company sought to defend.
Comments (0)