Musk's Grok Was Used to Help Target Iran Strikes, Pentagon Sworn Filing Confirms

A sworn declaration from the Pentagon's AI chief, filed in an unrelated environmental lawsuit, reveals that xAI's Grok was integrated into Palantir's Maven Smart Systems and helped the US deploy over 2,000 munitions against 2,000 distinct targets in 96 hours during 'Operation Epic Fury.' The disclosure lands alongside reports linking AI-assisted targeting to a strike on an Iranian girls' school that killed more than 175 people, mostly children.

TL;DR Cameron Stanley, the Pentagon's chief digital and AI officer, submitted a sworn declaration as part of the government's intervention in an environmental lawsuit against xAI's data centers. The filing confirms that Grok was integrated into Palantir's Maven Smart Systems (MSS) and helped deploy over 2,000 munitions against 2,000 distinct targets within 96 hours during "Operation Epic Fury." Separate reporting says AI-assisted targeting in the same period may have contributed to a strike on a girls' school in Minab, Iran, that killed at least 175 people, most of them children.

Elon Musk's chatbot has gone from answering questions on X to sitting inside the US military's targeting pipeline — and the public found out not through a Pentagon announcement or a congressional hearing, but through a court filing in a lawsuit that has nothing to do with combat operations. The NAACP sued to restrict xAI's Colossus data center in Memphis over alleged Clean Air Act violations, and the Trump administration intervened to argue the facility is now critical defense infrastructure.

A data center compared to a munitions plant

2,000
munitions deployed in 96 hours

2,000
distinct targets struck

3
enterprise AI providers certified for mission-critical use on Top-Secret networks

In his declaration, Stanley described xAI's computing capacity as "as foundational to our modern defense posture as traditional munitions production." He wrote that Grok is a core component of Maven Smart Systems, the Palantir-built platform that supports "targeting, intelligence, readiness, and recruitment," and that xAI is one of only three enterprise AI providers able to sustain "mission-critical operations" across both Secret and Top-Secret classified networks. According to the filing, MSS workflows enabled US forces to strike 2,000 distinct targets with 2,000 munitions in just 96 hours during Operation Epic Fury.

⚠️

A school was struck in the same window
Multiple outlets report that US military investigators believe forces relying on AI-assisted targeting were likely responsible for a strike on a girls' school in Minab, Iran, that killed at least 175 people, most of them children. Reporting indicates the school was selected based on outdated data processed through Palantir-linked systems that also incorporated Anthropic's Claude. Anthropic CEO Dario Amodei said last week the reported AI use in the Minab strike did not necessarily violate the company's "red lines," noting a human made the final call.

💡

Why this surfaced now
The disclosure came not from a planned Pentagon release but from a Department of Justice filing supporting xAI in litigation over Clean Air Act compliance at its Colossus facility. The government argued that shutting down or depriving the data center of power would impair Pentagon capabilities — and in making that argument, revealed Grok's military role.

Project Maven's long arc

Project Maven began in 2017 as a narrow effort to help analysts process drone footage. Google famously withdrew from the program in 2018 after employee backlash. The project has since expanded under Palantir, whose CEO Alex Karp has said the company's purpose is "to scare enemies, and on occasion, kill them."

Date	Event
2017	Project Maven launched to process drone footage
2018	Google withdraws after employee backlash
January 2026	Defense Secretary announces Grok's integration into Pentagon networks
Late February 2026	Military operations against Iran intensify
June 15, 2026	DOJ sworn filing discloses Grok's targeting role

xAI merged into SpaceX earlier this year, and Google and Anthropic have separately signed multibillion-dollar deals to rent compute capacity from Musk's company — meaning the implications of this disclosure extend well past a single firm into the AI industry's broader dependence on military-adjacent infrastructure.

This appears to be among the first court-confirmed disclosures that a commercial generative AI model was integrated directly into a lethal targeting pipeline.
AI-assisted targeting and a major civilian casualty event being reported in the same window puts real pressure on how AI providers' "red line" policies hold up in live military deployments.
Frontier AI companies now face simultaneous friction with the government on multiple fronts — export controls, supply-chain risk designations, and litigation — that could reshape how they approach defense contracts and model deployment generally.

Whether the full DOJ declaration becomes public, and whether Congress opens an inquiry, will likely determine how much technical detail outside researchers and watchdogs eventually get to review.

🔗