The US Department of Defence has stopped using Anthropic’s AI tools, an official told the BBC. The decision follows a February designation of Anthropic as a ‘supply chain risk’ on national security grounds.
What Happened
The Pentagon announced it has ceased all use of Anthropic products. This comes after Defence Secretary Pete Hegseth labelled the company a supply chain risk in February.
At the time, Anthropic refused to remove safety guardrails from its Claude AI models. The company cited concerns about mass surveillance and autonomous weapons.
According to multiple sources familiar with Pentagon operations, Anthropic’s Claude model was still being used in research, intelligence analysis, and military operations against Iran as recently as last week.
Key Facts
- Anthropic was the first advanced AI company to have its tools deployed in classified government work.
- Its tools were integrated into the Pentagon’s Maven Smart System, operated by Palantir.
- Defence officials and contractors reported that Claude was used to analyze satellite and drone imagery to identify potential military targets.
- Use of Claude supported the creation of one-page briefs passed up the chain of command.
- The Pentagon set a six-month deadline to phase out Anthropic, ending on 27 February — one day before the US and Israel launched operations against Iran.
Background: How the Tools Were Used
Palantir’s Maven Smart System is the Pentagon’s primary platform for organizing intelligence data. It processes visual data such as satellite and drone footage.
That data is fed into large language models like Claude. The AI helps analysts identify patterns and potential targets.
Lauren Kahn, a senior research analyst at Georgetown’s Center for Security and Emerging Technology, said removing deeply integrated tools like Claude is difficult. She noted that once AI is embedded in a system, it becomes hard to remove without significant effort.
Why It Matters
The Pentagon’s decision reflects broader tensions between AI capabilities and national security oversight.
Anthropic’s refusal to remove guardrails raised concerns about unchecked AI access. The company argued that such access could enable mass surveillance or autonomous weapons systems.
While the Pentagon has moved to stop using Anthropic, it has signed contracts with other AI firms. Google, xAI, and OpenAI are now being used in various military departments.
Relationships and Reactions
Despite the official designation, relations between the White House and Anthropic have improved.
Amodei expressed ongoing concerns about AI risks. He said, ‘How we address those risks is still under discussion,’ adding that ‘if we work with the president, we can win safely.’
Trump praised Amodei, calling him ‘great’ and ‘fantastic’ during the meeting.
Limitations and Open Questions
It is not clear why the Pentagon delayed full removal of Anthropic tools. The official statement came only after multiple sources reported continued use.
There is no public confirmation of how many systems or departments still used Claude before the phase-out. The full scope of the transition remains unverified.
Anthropic has sued the Trump administration, claiming the designation was ‘unprecedented and unlawful.’ A judge has previously ruled that the administration’s retaliation against Anthropic was illegal.
What to Watch Next
How the Pentagon manages AI integration moving forward will be closely watched. The decision may influence how other government agencies approach AI tools.
Any future use of AI in military operations will likely face increased scrutiny over safety guardrails and oversight mechanisms.
Sources & further reading
Featured image: Pre-colonial tools in National Museum of Ghana.jpg by Dieu-Donné Gameli, CC BY-SA 3.0, via Wikimedia Commons. Image source · License
