The United States Department of Defence has officially stopped using artificial intelligence tools provided by Anthropic, ending a controversial partnership that raised national security concerns. An official from the department confirmed the cessation Monday, months after the company was deemed a supply chain risk due to security issues.
National Security Concerns Prompt Action
Defence Secretary Pete Hegseth marked Anthropic as a supply chain risk back in February, proclaiming the Pentagon’s intention to cease using its products by late August. Nevertheless, sources indicated that Anthropic’s Claude model remained deployed within the department for activities such as research, intelligence gathering, and military actions against Iran, up until at least last week.
In response to these claims, the Pentagon confirmed it has ended the use of Anthropic products. Despite initial concerns that transitioning to new tools such as OpenAI would be challenging, the department has been gradually incorporating technologies from companies like Google, xAI, and OpenAI.
Integration Challenges with Maven Smart System
The Claude model was a critical component of the Pentagon’s Maven Smart System, which is operated by Palantir. This platform is used to organise intelligence and vast data, drawing on large language models for tasks like military target identification. Lauren Kahn, an analyst at Georgetown’s Center for Security and Emerging Technology, noted the complexities involved in removing integrated technology from such systems, contributing to the delay in phasing out Anthropic’s tools entirely.
Ongoing Legal and Diplomatic Developments
Anthropic’s refusal to remove safety features from its AI, citing issues such as mass surveillance and autonomous weapons, led to its designation as a “national security supply-chain risk”. Following this decision, which Anthropic described as “unprecedented and unlawful,” the company filed a lawsuit against the Trump administration to challenge the designation.
Despite these legal disputes, Anthropic’s relationship with the White House appears to be improving. Meetings between company CEO Dario Amodei and President Donald Trump have occurred, with discussions focusing on addressing AI risks. Amodei expressed optimism about collaborating with the administration to navigate these difficult issues.
The Pentagon’s latest actions underscore the evolving nature of partnerships in AI technology within government frameworks, as well as the challenges and policies surrounding the use of advanced technologies in national security contexts.
Reporting based on coverage first published by bbc.co.uk.