Elon Musk’s artificial intelligence tool Grok was used to identify targets to strike against Iran, a top defense official revealed Tuesday.
Grok, a generative artificial intelligence chatbot developed by Mr. Musk’s xAI, was used to deploy more than “2,000 munitions at 2,000 distinct targets within 96 hours during Operation Epic Fury, a testament to the greatly increased operational efficiency made possible by the Grok Gov Model,” according to a legal briefing.
The declaration — defending the trillionaire from a civil rights lawsuit that alleges xAI data centers are illegally polluting majority-Black communities’ air — seeks to protect the gas-fired turbines that power the $20 billion supercomputer that trains and upgrades its Grok models, some of which are used by the Defense Department.
Grok is among four AI models “currently capable of supporting national security applications,” and is also one of three products “equipped to sustain mission-critical operations” in top secret networks, said Cameron Stanley, the Pentagon’s chief digital and artificial intelligence officer.
Preserving the data center’s operational capacity is “a matter of paramount national security,” he wrote.
The Justice Department argued that the NAACP’s lawsuit, which accuses xAI of operating dozens of unpermitted turbines in violation of the Clean Air Act, “directly threatens our ongoing national security interests.”
It “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.”
Mr. Musk’s company says the turbines are temporary and mobile, and under state rules, this exempts the generators from air permits for up to one year.
The Defense Department relies on Grok Gov Model, tuned specifically for U.S. government specifications, to provide “critical support” to military operations, Mr. Stanley said, “targeting, intelligence, readiness, and recruitment.”
“And if xAl is hindered from continuing to improve and upgrade Grok, including the Grok Gov Model, DoW’s ability to meet its national security mission and keep pace with adversaries will be impaired,” he said in the briefing.
It’s unclear whether this tool was used in February’s missile strike on the Shajareh Tayyebeh elementary school in Minab, Iran, that killed roughly 150 people, most of whom were children.
Outside analysts have suggested that a combination of AI targeting and human error that failed to check whether target maps were up to date may have led to the mass casualties.
Satellite imagery analyzed by Amnesty International showed the facility had been partitioned off and converted to a school long before the strike. But according to a preliminary U.S. military investigation, the facility remained misclassified as an active military site in outdated defense databases — a targeting error the inquiry attributed to officers at U.S. Central Command who used obsolete data from the Defense Intelligence Agency.
Mr. Musky’s company signed an agreement with the Defense Department on Feb. 23, letting the Grok AI model be used in classified military systems. Operation Epic Fury started Feb. 28.
The U.S. military has used advanced AI tools, including Anthropic’s Claude and the Defense Department’s flagship artificial intelligence, the Maven Smart System, to generate targets.
But the federal government cut ties with Anthropic after the company refused the department’s demands to remove ethical safeguards barring its technology from being used for lethal autonomous weapons and the mass domestic surveillance of Americans.
The Trump administration labeled Anthropic a “supply chain risk to national security” — a designation historically reserved for foreign adversaries.
In turn, the department turned to its competitors, such as xAI.
Contact the author
Please read our comment policy before commenting.