Pentagon used Anthropic’s Claude AI in Iran attack despite ban: report

The US military reportedly used an artificial intelligence tool from Anthropic during airstrikes on Iran, despite an order from Donald Trump to stop federal use of the company’s technology, according to The Wall Street Journal.

The newspaper citing sources familiar with the matter confirmed the order to stop using the company’s systems came only hours before the mission began.

However, United States Central Command still used Anthropic’s Claude AI model to help with intelligence assessments, identifying potential targets and running simulated battle scenarios.

According to the report, the AI system has become embedded in some military workflows in the Middle East, even as tensions have grown between the company and the Pentagon over its use.

The dispute has been building for months. The Trump administration reportedly ordered agencies to stop working with Anthropic after defence officials labelled the technology a potential security and supply-chain risk during negotiations over how the military could access the company’s AI models.

The feud stems from Anthropic’s resistance to give Pentagon unrestricted use of its systems and its lobbying against parts of the administration’s AI policy.

The military has previously used Claude in other high-profile operations, including the capture of Venezuelan president Nicolás Maduro.