The kill chain surfaced in a pollution suit

A federal pollution case over air permits is now the public record showing that an Elon Musk chatbot was wired into a kill chain. The NAACP sued xAI in April under the Clean Air Act over the gas turbines powering its Memphis supercomputer — 57 of them, unpermitted, trailer-mounted, venting particulates and formaldehyde over a majority-Black neighborhood. To get the suit thrown out, the Justice Department filed a memorandum arguing that shutting the turbines down would harm 'national, economic, and energy security.'

Grok's government model 'enabled U.S. forces to deploy over 2,000 munitions to 2,000 distinct targets within 96 hours' during the Iran operation. — sworn declaration, US Department of Defense

The argument leans on a sworn declaration from the Pentagon's chief AI officer, who states that a government version of Grok runs inside the military's targeting system and was used during the recent Iran operation — by his account enabling US forces to direct more than 2,000 munitions at 2,000 targets in 96 hours. Grok is named as one of only four AI models cleared for the military's most classified networks. That admission is the government's own, made in a filing whose purpose is to keep a data center online, so the combat figure is its characterization rather than independently checked fact — but the naming itself is the news.

The shape of the thing is what matters: a private company's environmental liability and the use of a commercial AI model in combat are now legally fused, with national security offered as the shield against a clean-air statute. If it works, it is a template any AI data center can reach for — and the country learned a commercial chatbot was in the loop on live strikes not from a briefing but from a fight over diesel-grade turbine exhaust.