An intelligence report, compiled with the help of an AI chatbot, almost led to a Chinese cargo ship being intercepted by US forces in the Middle East, according to an exclusive CNN report published on Friday, August 18. The document, which circulated among US officials during the spring conflict with Iran, claimed the vessel was carrying components for a nuclear weapons program.
According to four sources interviewed by CNN, the military began planning the seizure operation. Two sources said armed troops were already preparing to board the ship, and fighter jets were in the air.
Just before the operation, a further review revealed that the report had been produced by an analyst from a special operations command with the help of a chatbot, which had incorrectly identified the ship’s cargo.
The analyst had consulted the tool about ship manifest data obtained from the Pacific Special Operations Command, based in Hawaii.
CNN was unable to determine the ship’s actual cargo or whether the chatbot was a commercial product or a government system. One source described the report as completely false and said it almost triggered a war.
AI is being integrated into almost all military and intelligence functions
The incident comes at a time when the Pentagon and US intelligence agencies are seeking to integrate AI into most of their activities, from analyzing large volumes of raw data and selecting targets to administrative tasks such as budgeting, logistics, and supply chains.
However, this effort is decentralized, according to officials interviewed by CNN. Different government sectors use different tools, with their own security rules and standards, and there is no single set of rules for validating the information generated by these systems.
As a result, the reliability of these tools varies significantly between the military and intelligence sectors.
Security analysts have warned for years that inaccurate or corrupted data can lead to catastrophic errors between nations. The incident involving the Chinese ship highlights a more immediate risk: that critical decisions may be made based on incorrect information produced by automated systems.
