Skip to content
NewsIncident

Pentagon's AI chief swore under oath that Grok guided 2,000 Iran strike targets in 96 hours

· by Pondero Newsdesk

The short version

A June 15 court filing unsealed details of Grok Gov Model's role in Operation Epic Fury, marking the first official U.S. government disclosure that a commercial AI lab's model was used in active combat targeting at scale.

Pentagon's AI chief swore under oath that Grok guided 2,000 Iran strike targets in 96 hours

A sworn court declaration by the Pentagon's chief digital and AI officer, filed June 15, 2026 in a Mississippi environmental lawsuit, contains the first official U.S. government acknowledgment that a major commercial AI lab's model ran inside an active combat targeting system. The detail surfaced not through a policy announcement or a congressional hearing but through a DOJ motion to dismiss an NAACP air-quality suit against an xAI data center.

What the declaration says

Cameron Stanley, the Pentagon's Chief Digital and AI Officer, stated under oath that a classified derivative called Grok Gov Model was integrated into Maven Smart System, the military's AI-enabled intelligence and targeting platform. Per Stanley's declaration, Maven Smart System "enabled U.S. forces to deploy over 2,000 munitions to 2,000 distinct targets within 96 hours" during Operation Epic Fury, per reporting by Wired and Yahoo News.

Grok Gov Model processed classified and open-source intelligence feeds from drones, satellites, and other sensors inside Maven's dashboard, flagged potential targets, and compressed planning timelines that had previously taken days into hours. Human operators then authorized each strike, per Stanley's framing. Whether that framing satisfies the Pentagon's stated "human-in-the-loop" policy is a question the filing does not resolve.

Stanley also described Grok Gov as one of four AI models approved for national security applications and one of three rated for mission-critical top-secret operations.

Why the filing made these disclosures

The context matters. The DOJ filed its June 15 motion not in a defense proceeding but to dismiss an NAACP lawsuit alleging Clean Air Act violations at xAI's Memphis-area data center, per TechCrunch's June 16 reporting. The NAACP alleged xAI expanded from 27 to 57 unpermitted gas turbines powering the facility. The DOJ's counter-argument: shutting down the turbines would threaten a data center "vital" to national security because it supports the Grok Gov Model pipeline that feeds Maven Smart System.

In practical terms, the government used classified operational details as a shield in an environmental case. The classified combat-targeting disclosures exist on the public docket because the DOJ chose to submit them there.

Why it matters for AI operators

The disclosure sets a concrete precedent on several fronts that people building or procuring AI systems should watch.

First, a commercial AI lab's model ran inside a live targeting system at scale. That is qualitatively different from using AI for logistics, translation, or after-action analysis. The debate over whether frontier models are ready for lethal decision-support just acquired a primary-source data point.

Second, the mechanism of disclosure matters as much as the disclosure itself. Classified AI military use can now surface through civil litigation, not just congressional oversight or inspector-general reports. Any company with a government AI contract should expect that operational details could appear in unrelated court filings.

Third, the environmental case may now be dismissed on national-security grounds, per TechCrunch. If that happens, a federal court will have ruled that AI infrastructure status can override Clean Air Act standing. That ruling would create a replicable template for shielding future AI data centers from local environmental regulation.

What to watch next

Two concrete milestones are worth tracking. The court's ruling on the NAACP's motion will determine whether AI infrastructure receives a functional national-security carve-out from environmental law. Separately, Congress has yet to respond with hearings on AI oversight in military targeting. The Armed Services committees have jurisdictional interest, and the Stanley declaration gives any member a primary-source hook for calling witnesses.

Sources