False AI Intelligence Report Nearly Triggered a U.S.-China Military Clash

A U.S. intelligence report built with help from AI nearly pushed the military into a move against a Chinese ship. CNN says the document circulated inside the U.S. military this spring, while the war with Iran was still underway, and claimed the vessel was carrying parts for a nuclear weapons program. The military started preparing to intercept it. Armed service members were getting ready to board, and aircraft were already in the air.

Then officials looked again. The report had been generated with help from a chatbot, and the machine had misread the ship’s cargo. CNN couldn’t establish what the ship was actually carrying. One source told the network the report was “entirely false” and said it had “almost started a war.” That is not dramatic language for effect. Against a Chinese ship, inside a tense military environment, it’s the kind of error that can turn into a real-world problem fast.

The chain apparently began when an analyst at U.S. Special Operations Command asked a chatbot about intelligence reporting tied to the ship’s manifest. That information came from Special Operations Command Pacific in Hawaii. It is still unclear whether the tool was a commercial product or one built by the U.S. government. A former senior U.S. official told CNN that the internal tools are mostly just commercial systems “wearing lipstick.” The chatbot mixed open-source material with classified signals intelligence, then the analyst used AI again to turn the results into a standard intelligence report — the kind decision-makers tend to trust.

Democratic lawmakers want an investigation. So do, apparently, the people who have to live with the result of “hallucinations” in a system built for speed. Sources told CNN that these errors have been showing up repeatedly as AI spreads through the intelligence community, and that younger analysts in particular are trusting the tools too much while pressure to move faster keeps rising. One source put it bluntly: “AI allows you to get to a bad idea faster.”

That problem is getting bigger, not smaller. The U.S. military and intelligence agencies are pushing AI into targeting, logistics, supply chains and raw intelligence analysis. Pete Hegseth has already unveiled an “Artificial Intelligence Acceleration Strategy,” and Donald Trump is now talking about an “AI Force” and an “AI Czar.” The real issue isn’t whether AI can be useful. It clearly can. It’s whether a system built to impress speed demons is being allowed to sit in the room when the consequences are missiles, ships and sleep-deprived humans under pressure.