United States Department of Defense (DoD) Halts Use of Anthropic AI Tools

[Financial News] The United States Department of Defense (DoD) has completely halted its use of AI tools developed by AI startup Anthropic, months after designating the company a national-security supply-chain risk.
On the 5th (local time), a DoD official told the British Broadcasting Corporation (BBC) in an interview, "The DoD has stopped using Anthropic products."
Previously, U.S. Defense Secretary Pete Hegseth announced in February that Anthropic would be placed on the national-security supply-chain risk list and that its use would be discontinued by the end of August. No clear reason was given for the delay beyond the original schedule.
According to a BBC report citing multiple sources familiar with the matter, Anthropic's AI model Claude continued to be used until recently for the DoD's research, analysis and intelligence gathering, as well as military operations against Iran. Former DoD officials and defense-industry figures said Claude played a key role in processing vast amounts of data.
An Anthropic spokesperson declined to comment on the complete suspension.
Anthropic had worked with the U.S. government and military authorities since 2024 and was the first advanced AI company to enter government agencies handling classified work. However, tensions surfaced early this year after the DoD demanded that Anthropic remove the safety guardrails applied to Claude and grant the military unrestricted access.
After Anthropic refused, citing concerns about mass surveillance and the development of autonomous weapons, Hegseth designated the company as a 'national security supply-chain risk' entity under a category applied to companies from hostile countries. Anthropic has since sued the Donald Trump administration to overturn the designation, calling it "an unprecedented and illegal action."
While the legal battle continued, the DoD signed contracts with other major IT companies, including Google Experience Center, xAI and OpenAI. Sources inside the U.S. military said the adoption of OpenAI tools had expanded at key DoD departments in recent months.
Experts cite Claude's deep integration with existing systems as the reason its removal was delayed. Claude had been built into the Maven Smart System (MSS), the DoD's core intelligence-integration platform operated by Palantir Technologies.
Lauren Kahn, a senior researcher at Georgetown University's Center for Security and Emerging Technology, explained, "Because Claude was deeply embedded in the Maven system, immediate removal would likely have been difficult. Extracting technology once it has been integrated into a system is extremely challenging."
In fact, DoD employees continued using Claude's latest models, including Claude Mythos, even as the conflict persisted. They reportedly fed visual data, including satellite imagery and drone footage, into Claude through MSS and used it for target identification and report writing.
Meanwhile, despite the conflict with the DoD, relations between Anthropic and the White House appear to be shifting. Anthropic CEO Dario Amodei visited the White House last week and met twice with President Donald Trump.
Amodei told reporters, "Discussions on how to address AI risks are still ongoing, but if we work with the president, we can win safely." President Trump responded by praising Amodei as "excellent."

[email protected] Yoon Jae-jun Reporter