Pentagon Confirms Grok AI Chatbot Used in Iran Strikes, Targeting 2,000 Munitions

The Pentagon's chief digital and artificial intelligence officer, Cameron Stanley, revealed in a sworn court filing that xAI's Grok chatbot was integrated with the U.S. military's Maven Smart Systems (MSS) to assist in targeting and deploying 2,000 munitions against 2,000 distinct targets within 96 hours during Operation Epic Fury, the Trump administration's designation for strikes against Iran. Stanley testified that the Grok Gov Model enabled "greatly increased operational efficiency" and provided tailored functionality for military planning workflows, report synthesis, predictive analytics for logistics, red-teaming analysis of adversary positioning, and personnel management.

The disclosure emerged in a legal filing submitted by the Justice Department as part of its defense of xAI against an NAACP lawsuit alleging the company operated unpermitted gas-burning turbines at its Colossus 2 data center near Memphis, Tennessee. The Pentagon argued that the data center is vital to national security missions and that disrupting it would threaten American national, economic, and energy security. According to reporting, Maven Smart Systems had previously relied on Anthropic's Claude before transitioning to Grok.

The filing marks the first formal acknowledgment that a commercially developed generative AI model played a direct operational role in kinetic targeting. While Stanley praised Grok's unique capabilities compared to other frontier AI models, the court filings do not disclose technical architecture, integration details, dataset provenance, or the extent of human oversight in the targeting process.

The revelation has prompted concern among Democratic lawmakers about AI deployment in warfare without adequate safeguards. Senator Kirsten Gillibrand of New York recently introduced legislation seeking to ban the use of large language models without human oversight in decisions involving the use of force, detention, or other high-consequence actions. Gillibrand warned that the Pentagon is "moving toward deploying incredibly powerful AI technology without commonsense guardrails in place, which could have catastrophic consequences." She emphasized that "the most critical decisions affecting our national security and the lives of our service members must always be made by human beings, not unaccountable machines."

Prior reporting has noted concerns about Grok's propensity to produce inaccurate outputs or problematic content, raising questions about the reliability of commercial AI systems in military applications. Industry observers have pointed to the need for greater transparency regarding model output traceability, audit trails for forensic review, and supplier governance in military AI procurement.