The US military has pressed Anthropic to loosen safety safeguards on its Claude AI model, escalating a dispute over how the company’s tools can be used for defense work. The pressure is framed as a high stakes test of whether an AI company can enforce limits on military applications of its own model.
The standoff came to light through The Guardian, which reported that Defense Secretary Pete Hegseth met with Anthropic CEO Dario Amodei and gave him until the end of the day on Friday to agree to Department of Defense terms or face penalties. The report said officials threatened to cancel a major contract and potentially designate Anthropic as a supply chain risk.
Anthropic has positioned itself as a safety forward AI company and has resisted use cases it says could involve mass surveillance or autonomous weapons without human input. The Guardian reported the Department of Defense has already integrated Claude into operations and is now signaling it may cut ties if Anthropic does not comply.
The Pentagon’s deadline raises the stakes for AI safeguards
The report described a weeks long dispute in which Anthropic pushed back against the Department of Defense terms, arguing its safeguards should not be adjusted to enable certain military uses. Defense officials, according to the report, viewed the company’s resistance as an obstacle to deployment.
The Department of Defense struck deals last July with major AI firms including Anthropic, Google, and OpenAI, with contracts potentially worth up to $200 million. It added that until this week, Claude was the only model cleared for use in the military’s classified systems, amid Epstein brother interview fallout.
That changed on Monday when the Department of Defense signed a new deal allowing personnel to use Elon Musk’s xAI chatbot in classified systems. The report noted xAI faced recent backlash for producing nonconsensual sexualized images of children.
Both xAI and OpenAI have agreed to the government’s terms, and that one defense official said OpenAI has allowed its model to be used for all lawful purposes. That left Anthropic as the remaining holdout in the negotiations described in the report.
The US military used Claude last month to help with the capture of Venezuelan leader Nicolás Maduro, and said President Trump’s administration has pushed to integrate AI into the military. The report described public comments from Emil Michael, the Pentagon’s chief technology officer, urging Anthropic to agree to the Department of Defense terms and arguing guardrails should be tuned for military use cases so long as they are lawful.
Amodei has supported stronger AI regulation and Anthropic backs a political action committee advocating greater safeguards for artificial intelligence. The report also said Amodei opposed Trump during the 2024 campaign and that Anthropic has hired several former Biden staffers, developments it linked to political fallout in the company’s fundraising.
The Pentagon is investing heavily in AI enabled technologies including drones and automated targeting systems, intensifying debates over how much decision making should be delegated to AI in warfare. That push is unfolding alongside attention on cyanobacteria desertification program. It pointed to semiautonomous drones operating in the conflict in Ukraine without direct human control as part of that broader context.
Published: Feb 25, 2026 06:15 am