Pentagon Says Grok Aided 2,000 Strikes Against Iran

The Justice Department asked a federal judge on June 15 to dismiss an NAACP Clean Air Act lawsuit against xAI — and Pentagon AI chief Cameron Stanley swore under oath that Grok's Gov Model helped U.S. forces deploy 2,000 munitions against 2,000 Iranian targets in 96 hours during Operation Epic Fury. On June 30, environmental lawyers called that argument a license for lawbreaking if the White House says "national security."

What's new:

• In NAACP v. xAI Corp. (N.D. Miss., No. 3:26-cv-00074), the DOJ moved to intervene and dismiss, arguing citizen suits cannot proceed when the executive branch objects.
• Stanley's declaration: Grok Gov Model integration into Maven Smart Systems enabled the Iran strike tempo — "a testament to the greatly increased operational efficiency made possible by the Grok Gov Model."
• SELC, representing the NAACP, responded June 30: DOJ "never disputes that xAI is pumping out unlawful and harmful pollution" — it just argues that shouldn't matter.

The battlefield disclosure:

• Stanley: Grok's "continued operation and availability is a matter of paramount national security."
• Grok Gov Model is one of four frontier models "currently capable of supporting national security applications" and one of three equipped for mission-critical work on Secret and Top Secret networks, per the filing.
• MSS users process nearly two billion tokens daily on DoD's top-secret network — roughly 1.5 billion words, per Stanley.
• Stanley did not detail human-in-the-loop controls, error rates, or how Grok outputs fed final strike authority.

The pollution fight:

• NAACP sued in April alleging xAI and MZX Tech operated gas turbines at the Colossus Gas Plant in Southaven, Mississippi — powering Colossus data centers that run Grok — without Clean Air Act permits.
• NAACP's June 12 filing: unpermitted turbines rose from 27 to 57 by mid-May, with two more planned.
• Plaintiffs seek injunction, civil penalties up to $124,426 per day, and fees — alleging tenfold excess nitrogen oxide emissions without best-available controls.
• Mississippi DEQ told DOJ the turbines are "mobile sources" not requiring permits; Gov. Tate Reeves approved permanent turbine construction in March.

DOJ's legal theory:

• "The NAACP's attempt to cut off the power that supports Grok also threatens national security because… Grok provides critical support for the Department of War's military operations."
• DOJ: Clean Air Act citizen suits cannot "commandeer the federal enforcement machinery" when the United States chooses to forgo enforcement.
• Associate Attorney General Stanley Woodward: enforcement "falls to the executive branch, not private interest groups."
• Principal Deputy ENRD chief Adam Gustafson: DOJ "will not sit idly by while private organizations use environmental laws to undermine our national security."

Why this matters beyond Mississippi:

• xAI signed a DoD agreement in February 2026; last year the Pentagon awarded xAI up to $200 million for agentic AI workflows.
• SpaceX now owns xAI — so Grok's war role lands inside the week's hottest public listing.
• If DOJ wins, any contractor invoking federal AI priority could argue citizen enforcement is preempted — a precedent SELC calls "an unprecedented attack on the public's ability to defend themselves from illegal pollution."

The pushback:

• Earthjustice enforcement director Laura Thoms: "This isn't about national security. It's a desperate attempt to protect wealthy corporations like xAI from obeying our bedrock environmental laws."
• SELC litigation director Kym Meyer: the motion contradicts "decades of well-established legal precedent."
• Army Times noted Stanley never explained how operators actually used Grok during combat — only that it supported targeting workflows.

Convina's view: Washington spent June gatekeeping Anthropic and OpenAI behind guest lists while quietly running Grok on classified targeting networks — then used that deployment as a pollution defense for unpermitted gas turbines next to majority-Black communities. That is not AI safety policy. It is AI as sovereign immunity. The disclosure that Grok touched 2,000 Iran targets in four days is the real product spec sheet — and nobody outside the Pentagon has seen the eval data, the override switches, or the liability chain when the model is wrong. Until Congress demands that chain, "national security" will keep meaning whatever keeps the turbines spinning.