A wildly inaccurate U.S. military intelligence report produced with artificial intelligence almost started World War III, CNN reported Friday.
In the spring, U.S. military officials launched an operation to intercept a Chinese ship they believed was transporting nuclear weapons materials in the Middle East. Service members prepared to board ships, and military aircraft were already in the air, multiple sources told CNN.
That’s when officials realized that the threat wasn’t real. It was AI-generated.
The claim had stemmed from a report that was produced with the help of a chatbot, but the AI program had inaccurately identified the materials the ship was carrying. CNN was unable to identify what the ship was actually carrying. It was not clear whether the chatbot was commercially available or the government’s own knock-off version.
One of the sources told CNN that the report was “entirely false” and had “almost started a war.”
The Trump administration has been eager to embrace the AI industry—a risky bet considering the technology’s tendencies to hallucinate answers to queries.
In July 2025, the government awarded Google a massive $200 million contract to support AI solutions at the Defense Department. Six months later, Defense Secretary Pete Hegseth rolled out GenAi.Mil, the new “American-made” AI platform that will allow military members to “conduct deep research, format documents, and even analyze video or imagery at unprecedented speed”—and all without using their brains. It’s not clear whether this is the tool that falsely identified materials to make nuclear weapons.
In January 2026, Hegseth doubled down on the technology with an “AI Acceleration Strategy” at the DOD. A few months later, he promised that the U.S. military was using “every tool of AI” in the war against Iran.