The US Department of Defence has stopped using Anthropic’s artificial intelligence tools after designating the company a national-security supply-chain risk, a Pentagon official told the BBC. Defence Secretary Pete Hegseth issued the designation in February and ordered Anthropic’s products to be phased out by late August, though the reason for the delay is unclear.
Sources familiar with Pentagon operations said Anthropic’s Claude models remained in use as recently as the previous week for research, intelligence gathering, analysis and military operations involving Iran. Claude was reportedly integrated into Palantir’s Maven Smart System, the Pentagon’s main platform for organising intelligence, including satellite imagery and drone footage. Analysts used the system and large language models to identify possible targets and prepare briefings for senior officials.
The Pentagon had asked Anthropic to remove safety restrictions and provide the military with unrestricted access. Anthropic refused, citing concerns about mass surveillance and autonomous weapons. Hegseth then labelled the company a supply-chain risk, a move Anthropic called unlawful and is challenging in court.
Experts said Claude’s deep integration into Maven made rapid removal difficult, demonstrating that AI systems are not easily interchangeable once embedded in government infrastructure. Meanwhile, the Pentagon has expanded contracts with Google, OpenAI and xAI, with OpenAI tools reportedly gaining wider military adoption.
Despite the dispute, relations between Anthropic chief executive Dario Amodei and President Donald Trump appear to be improving following recent White House meetings, although Amodei continues to warn about AI risks.
Article timeline
Build this article’s timeline
Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.
Pete Hegseth, Dario Amodei, Donald Trump, and United States Department of Defense
🤖 AI prefill
📰What
The US Department of Defence has stopped using Anthropic’s artificial intelligence tools after designating the company a national-security supply-chain risk, a Pentagon official told the BBC.
🤖 AI prefill
📅WhenWhen did this happen, or what is the timeframe?
📍Where
San Francisco, Washington, and Iran
🤖 AI prefill
💡Why
Centers on Pentagon cessation of Anthropic AI tools, Anthropic national security supply-chain risk designation, and Claude AI model
🤖 AI prefill
Login to answer and earn XP for each gap you fill.
Pentagon Anthropic BanPentagon, Anthropic, Claude, blacklisting, supply-chain riskMilitary AI Operationsintelligence, military operations, target identification, satellite imagery, drone footageAI Safety Disputesafety guardrails, mass surveillance, autonomous weapons, biological weapons, AI risksLegal Political Conflictlawsuit, Trump administration, Pete Hegseth, Dario Amodei, national securityDefense AI PlatformsMaven Smart System, Palantir, OpenAI, Google, xAI
Key Phrases:Pentagon cessation of Anthropic AI tools99%Anthropic national security supply-chain risk designation98%Claude AI model96%Maven Smart System96%Palantir intelligence platform93%military intelligence gathering and analysis91%AI safety guardrails90%autonomous weapons89%mass surveillance87%large language models86%
Comments
Login to view and post comments.