Revealed: The Pentagon Used Elon Musk’s AI in Strikes on Iran

Grok, the artificial intelligence system developed by Elon Musk’s xAI, was used by the U.S. military as part of strikes carried out against targets in Iran, according to a U.S. Department of Justice legal filing revealed this week. The information was reported in foreign media outlets and was included in an affidavit submitted to the court as part of a legal battle surrounding the company’s server farm.

The document, filed on June 15, was intended to defend xAI’s operations against an environmental lawsuit accusing the company of running dozens of gas turbines without the proper permits. As part of that defense, the Department of Justice argued that the lawsuit “threatens the national, economic, and energy security of the United States by seeking to cut off the electricity supply to artificial intelligence innovation that supports the military activities of the Department of War.”

To reinforce that argument, federal prosecutors presented testimony from the Pentagon’s head of artificial intelligence, Cameron Stanley, who stated under oath that Grok had already been integrated into Project Maven, the U.S. military’s AI-based operational support program for identifying targets and streamlining strike processes.

Stanley said Musk’s technology contributed to the “significantly increased operational efficiency made possible through the Grok Gov Model.”

According to the reports, the government version of Grok was integrated as an intelligence support tool for locating and analyzing potential strike targets as part of the war against Iran. The Pentagon also described xAI’s computing infrastructure as a “vital strategic asset for maintaining the technological advantage.”

The lawsuit against xAI was filed by the civil rights organization NAACP, which argues that the company is operating dozens of polluting turbines without the required approvals, harming neighborhoods whose residents are mostly Black. The company, for its part, argues that these are temporary and mobile facilities that are not subject to the same regulatory requirements.

The disclosure joins a broader controversy over the integration of artificial intelligence into military operations. According to the report, at the end of February the U.S. government ended its contracts with Anthropic after the company refused to allow its systems to be used for fully automated strikes or mass surveillance of American citizens. Following that, the Pentagon turned to other companies in the field, including Google, OpenAI, and xAI.

The move is also drawing criticism from within the tech industry. More than 600 Google employees previously called for the company to avoid making AI capabilities available to the military for classified missions, while other voices have warned about the risks involved in relying on AI systems for operational decision-making.

The revelation is already causing a political uproar in Washington. Democratic lawmakers are advancing initiatives to limit the military use of artificial intelligence, in part because of claims that mistakes in target identification or the use of outdated information could lead to harm to civilians during military strikes.