U.S. military forces, according to media reports, used Claude—the language model developed by Anthropic—while preparing a strike on Iran, despite an order issued by Donald Trump just hours before the operation began to immediately sever all ties with the company and its AI tools.
The use of Claude during a large-scale joint U.S.–Israeli operation against Iran, which began on Saturday, was reported by The Wall Street Journal and Axios. The episode underscored how difficult it has become for the Pentagon to disengage from powerful AI systems once they are deeply embedded in military operations.
According to the Journal, commanders used these tools for intelligence tasks, target selection, and the simulation of combat scenarios.
On Friday, February 27, just hours before strikes on Iran began, Trump ordered all federal agencies to immediately stop using Claude. In a post on Truth Social, he described Anthropic as a “radically left-wing AI company run by people who have no understanding whatsoever of how the real world works.”
The confrontation escalated after Claude was used by U.S. military forces in a January operation to capture Venezuelan President Nicolas Maduro. Anthropic objected, citing its terms of use, which prohibit employing the model for violent actions, weapons development, and surveillance activities.
Since then, relations between Trump, the Pentagon, and the company have steadily deteriorated. In an extended statement on X on Friday, Defense Secretary Pete Hegseth accused Anthropic of “arrogance and betrayal,” stressing that “the U.S. military will never be held hostage to the ideological whims of Big Tech.”
Hegseth demanded full and unrestricted access to all Anthropic AI models for any lawful purpose. At the same time, he acknowledged that rapidly disconnecting military systems from these tools would be extremely difficult given how widely they are used. According to him, the company will continue providing services “for a period of no more than six months to ensure a smooth transition to a higher-quality and more patriotic service.”
After the break with Anthropic, the resulting gap was filled by OpenAI. The company’s chief executive, Sam Altman, said an agreement had been reached with the Pentagon to deploy the company’s AI tools, including ChatGPT, within a closed military network.