The US Department of Defence has ceased using Anthropic's AI tools, as confirmed by an official to the BBC. This decision follows the Pentagon's designation of the company as a supply chain risk related to national security. Secretary of Defence Pete Hegseth labeled Anthropic a supply chain risk in February, with plans to stop using its products by late August. However, reports indicate that Anthropic's Claude model was still in use as recently as last week for various military operations, including intelligence gathering against Iran.
The Pentagon's official statement on the cessation of Anthropic products was made on Monday. Anthropic had been utilized by the US government and military since 2024, being the first advanced AI company to have its tools deployed in classified government work. Earlier this year, the Pentagon requested Anthropic to remove safety guardrails from its Claude model, seeking unrestricted access to its AI products. Anthropic declined this request, citing concerns over mass surveillance and autonomous weapons, leading to Hegseth's designation of the company as a national security supply-chain risk.
Anthropic has described the Pentagon's actions as "unprecedented and unlawful" and has initiated legal proceedings against the Trump administration to challenge the designation. As the legal situation unfolds, the Pentagon has engaged with other technology companies, including Google, xAI, and OpenAI, with OpenAI's tools seeing increased adoption in military departments.
The Pentagon's statement was prompted by inquiries regarding the continued use of Anthropic's tools, despite previous assertions that they would be phased out. Sources indicate that Claude was integrated into the Maven Smart System, a key data platform operated by Palantir, complicating its removal. Lauren Kahn, a senior research analyst at Georgetown's Center for Security and Emerging Technology, noted that the integration of such tools can make their removal challenging.
Hegseth's initial order to phase out Anthropic came shortly before a US-Israel operation against Iran. Reports suggest that Pentagon employees continued to utilize Claude for intelligence tasks throughout the controversy. Maven is also used by some US allies for intelligence sharing.
In recent developments, Trump met with Anthropic's Dario Amodei at the White House, where Amodei expressed ongoing concerns about AI risks. Trump praised Amodei during the meeting, indicating a potential thaw in relations between the company and the administration.