The role of artificial intelligence in military operations and the responsibilities of the companies that develop it are under renewed scrutiny this week, following a series of developments involving two of the industry's most prominent players.

Anthropic's Claude AI was allegedly used by the Pentagon during the joint US-Israel bombardment on Iran on 28 February. While the exact role of the AI in the operation has not been disclosed, the Wall Street Journal reported that US military command used the tools for intelligence gathering, target selection and battlefield simulations. Neither the Pentagon nor Anthropic responded to Reuters’ requests for comment.

Claude AI was allegedly used just hours after President Trump instructed federal agencies to halt its use, following the company’s refusal to grant the Pentagon unrestricted access to its systems for autonomous weapons targeting and mass domestic surveillance.

Reports suggest that the gap left by Anthropic's withdrawal has partly been filled by OpenAI, which has signed a deal with the US Department of War to use its tools on classified networks. In a statement, OpenAI said that its agreement includes explicit prohibitions on domestic surveillance of American citizens, directing autonomous weapons and high stakes automated decision making and “has more guardrails than any previous agreement for classified AI deployments”.

OpenAI’s decision to enter into the agreement has triggered a strong user backlash, with reports indicating that uninstalls surged 295 per cent in the USA.