The US military utilized a version of Elon Musk's Grok artificial intelligence to orchestrate rapid-strike operations during the war with Iran, a senior Pentagon official revealed in a court declaration.

“MSS (Maven Smart Systems) frontier workflows enabled US forces to deploy over 2,000 munitions to 2,000 distinct targets within 96 hours during Operation Epic Fury,” said Cameron Stanley, chief digital and artificial intelligence officer for the Pentagon, in a sworn statement submitted on Monday in a federal case in the Southern state of Mississippi.

He characterized the achievement as a testament to the "greatly increased operational efficiency" provided by the Grok Gov model.

Maven Smart System is an AI-supported command and data analysis platform developed by the Pentagon and software company Palantir.

Stanley gave the statement as part of a federal case involving an environmental lawsuit that seeks to shut down the gas-fired turbines powering xAI’s Colossus 2 data center. He argued that any disruption to the facility would directly threaten national security interests, as the military relies on the supercomputer to train and upgrade its AI capabilities.

He compared the importance of massive data centers to traditional ammunition manufacturing. “The critical ability to field its data facilities at massive scale is as foundational to our modern defense posture as traditional munitions production,” Stanley asserted. He said users on a Pentagon network use some 1.5 billion words daily to manage logistics, predictive analytics, and military planning.

The Pentagon's reliance on xAI follows the termination of its partnership with rival company Anthropic, which refused to permit its Claude AI model to be used for fully automated strikes or domestic mass surveillance.

The Washington Post reported in March that the Pentagon used the Claude model and the Maven Smart System while planning attacks on Iran.

The report said the Maven system used advanced AI technology to identify potential targets in Iran and provide their precise locations.

It also claimed that the system was used in an operation this January to capture Venezuelan President Nicolas Maduro through a military operation.

Versions of Grok have previously been the subject of controversy, including last year when the Grok chatbot praised Hitler and later when it allowed users to alter images of individuals, including minors, to show them in revealing clothing or in sexually suggestive contexts. (AA)

Update news