Ships pass through the Strait of Hormuz as seen from Musandam, Oman. [Reuters]
Ships pass through the Strait of Hormuz as seen from Musandam, Oman. [Reuters]

The US military prepared an operation against a Chinese vessel based on a fabricated intelligence report generated with AI assistance, it has emerged.

According to CNN, an intelligence report circulated within the US military this past spring, during the Iran war, claiming a Chinese ship in the Middle East was transporting components related to a nuclear weapons program.

Multiple sources said the military immediately began preparing an operation to intercept the vessel based on the report. Armed US troops were ready to board the ship, and military aircraft had already been scrambled.

The operation was halted just before execution, however, after officials reviewing the report discovered it had been written by a Special Operations Command analyst with AI assistance — and that the AI chatbot had misidentified the cargo aboard the vessel.

Had the operation proceeded without that additional review, it could easily have escalated into an armed conflict between the United States and China.

One source said the report was "completely false" and that it had nearly started a war.

CNN said the incident illustrates the danger that AI errors could lead to real military miscalculations, amid the rapid expansion of AI use across the Ministry of National Defense and intelligence agencies.

Defense Secretary Pete Hegseth announced an "AI acceleration strategy" in January and has been expanding AI adoption across military functions, including intelligence analysis and targeting. The trend has moved beyond using AI as an administrative or analytical tool, extending its role into combat planning and targeting decisions.

"There is clearly a growing trend of using AI in targeting," one source said. "There are no real guidelines on how to prevent civilian casualties or friendly-fire incidents."

Some are also calling for safeguards similar to those applied to nuclear weapons to address the risk of AI miscalculation in military settings. The concern is that if military AI assesses threats or supports strike decisions based on faulty information, humans may have very little time to catch the error and intervene before a miscalculation leads to actual conflict.

Reuters recently reported that US and Chinese experts have proposed establishing an emergency communication channel between the two countries to guard against AI malfunctions or mutual misidentification of military systems during a crisis, and reaching advance agreements on the operational scope of AI systems and the principle of final human approval.


balme@heraldcorp.com