Grok AI was used in US military strikes against Iran, government confirms

Elon Musk’s AI chatbot has gone from answering questions on X to helping the US military pick targets in Iran. A Department of Justice legal brief dated June 15, 2026, confirmed that xAI’s Grok was integrated into the Pentagon’s Project Maven and played a direct role in combat operations.

Pentagon AI chief Cameron Stanley testified under oath that a government-focused version of Grok, called the “Grok Gov Model,” was deployed within Maven Smart Systems for AI-assisted targeting. During a military campaign dubbed Operation Epic Fury, more than 2,000 munitions were deployed against distinct targets within just 96 hours.

From chatbot to combat tool

Defense Secretary Pete Hegseth first announced Grok’s integration into Pentagon networks back in January 2026. By February, it was apparently doing far more than administrative work.

According to the testimony, Grok predicted a significant strike date of February 28, 2026, on Iran. That prediction aligned with subsequent military actions, suggesting the AI wasn’t just processing data but actively shaping operational timelines.

Project Maven has been around since 2017, originally designed to use machine learning to analyze drone footage. Google famously walked away from the project after employee backlash.

The Grok Gov Model processed intelligence data to rapidly identify and prioritize threats during a period of escalating tensions with Iran in early 2026.

The details surfaced in legal documentation connected to an environmental lawsuit against xAI, not through a planned Pentagon announcement or congressional hearing.

The scale of Operation Epic Fury

Deploying over 2,000 munitions against distinct targets in 96 hours is an extraordinary tempo of military operations. That’s roughly one target struck every three minutes, around the clock, for four straight days.

The AI’s role wasn’t limited to a single function. It handled threat identification, target prioritization, and predictive analysis, essentially touching every step of the kill chain before a human commander made the final call.

Stanley’s testimony referenced “AI-assisted targeting,” which implies human oversight remained in the loop. But the speed at which targets were being struck suggests the degree of human review per target may have been minimal.

What this means for investors and the AI industry

Google abandoned Project Maven in 2018 precisely because it feared the backlash would hurt recruitment and public trust. xAI doesn’t appear to share that concern, but the company still operates in a commercial market where public perception matters.

The European Union has already been moving toward stricter AI governance frameworks. Confirmation that a Western commercial AI product was used in military strikes could accelerate those efforts and create compliance headaches for any AI company operating across borders.

The legal proceedings that surfaced this information are ongoing. An environmental lawsuit producing revelations about military AI targeting is the kind of procedural accident that tends to generate follow-up investigations, congressional inquiries, and potentially new disclosure requirements for AI companies working with the defense establishment.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.