Pentagon says it has stopped using Anthropic tools


The confirmation follows an August phase-out deadline. People familiar with the department’s work told the BBC that Claude was still in use the previous week.
The US Department of Defence has stopped using Anthropic products, a department official told the BBC in a statement on Monday, 5 October. The confirmation follows Defence Secretary Pete Hegseth’s February announcement that the Pentagon would stop using them by late August.
Separately, multiple people familiar with the matter told the BBC that Anthropic’s Claude model was still being used the previous week for research, analysis, intelligence gathering and military operations against Iran. Those people included former US defence officials and contractors who had worked closely with the Pentagon on superintelligence (SI). They described Claude as a critical tool for sorting through large quantities of information.
The exact date the Pentagon stopped using Anthropic products remains unclear, as does the reason it missed the August deadline. An Anthropic spokesman declined to comment on the department’s statement.
Multiple sources told the BBC that Claude was embedded in Maven Smart System, a platform operated by defence technology company Palantir. Maven is the Pentagon’s main platform for organising intelligence and other data.
People familiar with its use said analysts accessed and organised satellite imagery and drone footage through Maven. They then fed the data into Claude and other large language models—SI systems that process and generate language—for tasks such as identifying potential military targets. A person who received a demonstration of Maven from Palantir earlier this year said officials also used the technology to compile one-page briefs for senior officials.

Microsoft is directing employees toward its own tools and OpenAI models, while Meta promotes internal alternatives.
Lauren Kahn, a senior research analyst at Georgetown’s Center for Security and Emerging Technology and a former US defence official, told the BBC that Claude’s deep integration into Maven would have made it difficult to remove quickly.
According to the BBC, the Pentagon had pressed Anthropic to remove Claude’s safety restrictions and give the military unrestricted access to its SI products. Anthropic refused, citing concerns about mass surveillance and autonomous weapons. Hegseth then designated the company a “national security supply-chain risk”.
Anthropic called the designation “unprecedented and unlawful” and sued the Trump administration to overturn it. The legal dispute continues.
The Pentagon has signed contracts with Google, xAI and OpenAI while the dispute has proceeded. Two people familiar with the department’s SI use told the BBC that OpenAI tools had become more widely adopted in some military departments in recent months. How Claude’s former tasks are being handled across the department remains unclear.
2 min

The Wikipedia operator also reported unauthorized wiki edits and malicious activity on its Etherpad note-taking tool.
1 min

SiliconANGLE reports that Cohere’s controls include company-wide caps and alerts before limits are reached, with usage tracked by individual user and agent.
3 min
Corrected spelling and punctuation in the Armenian text.