The United States government has inadvertently disclosed that a derivative of Elon Musk’s Grok artificial intelligence is being used by the Pentagon to assist in military strikes against Iran. The revelation emerged in a 15 June legal filing by the Department of Justice, which sought to dismiss an environmental lawsuit threatening the gas turbines powering xAI’s vast data centre in the American South. The brief argued that the lawsuit “threatens American national, economic, and energy security” by jeopardising the AI infrastructure that supports the Department of War’s operations. To bolster this claim, federal prosecutors submitted sworn testimony from Cameron Stanley, the Pentagon’s AI chief, confirming that a “Grok Gov Model” is already integrated into Project Maven, the military’s controversial AI-assisted targeting programme.

Viewed from Washington, the intervention underscores the Trump administration’s willingness to shield Musk’s commercial ventures by framing them as essential to national defence. The lawsuit, brought by the NAACP and other groups, alleges that xAI illegally operated dozens of natural gas turbines without permits, creating health risks for communities near the Mississippi-Tennessee border. The administration’s legal manoeuvre effectively elevates corporate protection to a matter of national security, a move that critics say prioritises military expediency over environmental compliance and public health.

From Tehran, the disclosure is likely to deepen an already volatile confrontation. Iranian media outlets aligned with the Islamic Revolutionary Guard Corps had previously warned that Musk’s entire business empire—including Starlink, SpaceX, and the X platform—was on a military “hit list.” Analysts in London note the irony that a private AI tool, developed by a figure who has oscillated between government advisory roles and entrepreneurial pursuits, is now embedded in kinetic warfare. European capitals, long uneasy with Project Maven’s blurring of public and private interests, are watching closely as the programme shifts from Anthropic’s Claude to Grok, raising fresh questions about accountability in AI-assisted lethal operations.

The case is likely to set a significant precedent, potentially enabling other technology firms with government contracts to invoke national security in sidestepping environmental regulations. It also sharpens scrutiny of Musk’s dual role as both a defence contractor and a political confidant of the president, recently seen together at a summit with Xi Jinping in Beijing. As the boundaries between commercial AI, military targeting, and geopolitical strategy continue to dissolve, the Memphis data centre stands as a physical emblem of an uncomfortable convergence: the same energy infrastructure that powers a chatbot capable of generating viral content is now, by the government’s own admission, helping to select targets in the Middle East.