How Grok AI Helped the US Military Strike 2,000 Targets in Iran Within 96 Hours

How the Disclosure Came Out in a Courtroom

The details of Grok's battlefield deployment were not announced via press release. They emerged in a June 15, 2026 legal brief filed by the DOJ in a Mississippi federal court . The government was intervening in an environmental lawsuit against xAI brought by the NAACP, and used Stanley's sworn testimony to argue that any disruption to xAI's data centers would damage paramount national security interests .

The brief argued that the lawsuit “threatens American national, economic, and energy security by seeking to shut off the power supply for artificial intelligence innovation that supports the Department of War’s military operations” . It was in this context—defending the operation of unpermitted gas turbines—that the full scope of Grok's integration into Project Maven was laid bare.

Project Maven and the Rapid Pivot from Anthropic to Grok

Project Maven, originally launched to process drone surveillance footage, has evolved into a broader targeting and intelligence platform known as Maven Smart Systems (MSS), developed by Palantir . In early 2026, the Pentagon's classified AI partner was Anthropic, the developer of the Claude model .

By July 2025, xAI had already secured a $200 million contract with the Pentagon for its "Grok for Government" initiative, granting government agencies access to its Grok 4 model . After a compliance dispute with Anthropic, Defense Secretary Pete Hegseth issued an ultimatum in February 2026, threatening to invoke the Defense Production Act and labeling Anthropic a supply-chain risk if it did not agree to military terms .

On January 13, 2026, Hegseth publicly announced that Grok would be integrated into the Pentagon's Google-powered generative AI network, effectively positioning Grok to replace Anthropic as the primary AI model for classified military operations . Senators later raised concerns about Grok’s history of generating antisemitic content, but the integration proceeded .

The Environmental Lawsuit Against xAI's Data Center

The lawsuit that triggered Grok's disclosure was filed in April 2026 by the NAACP, represented by Earthjustice and the Southern Environmental Law Center . It accuses xAI and its subsidiary MZX Tech of violating the Clean Air Act by operating nearly 50 unpermitted natural gas turbines on flatbed trailers at its Colossus 2 data center in Southaven, Mississippi, a facility across the border from Memphis, Tennessee .

Mississippi law classifies the turbines as "mobile" because they sit on trailers, enabling xAI to avoid standard air pollution permits for up to a year . The lawsuit alleges that the turbines are located near homes, schools, and churches, and that their emissions—potentially over 1,700 tons of smog-forming nitrogen oxides annually—pose disproportionate health risks to surrounding Black communities . The EPA had previously issued a rule in January 2026 stating that such “temporary” turbines must still be regulated as stationary sources, directly contradicting xAI's permitting stance .

In May 2026, the NAACP filed for a preliminary injunction to halt the pollution .

The Government's National Security Argument

In its June 15 brief, the DOJ argued that the court should dismiss the lawsuit outright. The key national security claims included:

• Only four AI models—including Grok—support mission-critical operations across Secret and Top-Secret networks, and Grok is one of just three products equipped for Top-Secret environments .
• Stanley's declaration framed xAI's data centers as “battlefield infrastructure of paramount national security importance” .
• Disrupting power to xAI's facilities would directly harm ongoing military operations, most critically the Iran campaign .
• The government argued that xAI is permitted to run the turbines without standard permits because of overriding national security needs .

Timeline of Events

The revelation has reignited a global conversation about the use of commercial AI for lethal targeting, the environmental footprint of AI infrastructure, and how far national security claims can stretch in civil litigation . It also raises questions about the oversight of AI models that earlier in 2026 were involved in deepfake abuse scandals and generated antisemitic content, yet now sit on Top-Secret Pentagon networks .