Chatbot hallucination nearly triggered a US military operation

CNN reported that an analyst at US Special Operations Command Pacific used an AI chatbot to help assess the cargo of a Chinese-flagged vessel in the Middle East during the spring 2026 Iran war period, and that the chatbot, combining open-source material with classified US signals intelligence, produced a false conclusion that the ship was carrying components for nuclear weapons. The analyst then used the same tool to format that fabricated finding into an official-looking intelligence report, which circulated through command channels.

According to CNN’s sources, armed personnel had prepared to board the vessel and military aircraft were already airborne before officials traced the claim to its AI origin and aborted the operation. One source described the underlying report as “entirely false”; another told CNN the episode “almost started a war.” CNN did not name the specific chatbot used, and neither TechCrunch’s report nor other outlets covering the story identified the underlying model or its developer.

The incident, which CNN said occurred that spring, only became public when its report was published in September 2026. It was cited afterward as a concrete example of a risk that had mostly been discussed hypothetically — a fabricated AI output entering a classified intelligence process and nearly triggering an armed response, rather than being caught at the analysis stage.