"Almost Started a War"...U.S. Military Nearly Attacked Chinese Vessel Based on Fake AI-Generated Intelligence

The U.S. military reportedly prepared a military operation against a Chinese vessel based on misinformation generated by artificial intelligence (AI), but detected the error shortly before execution and called off the operation.

According to CNN on the 18th (local time), during the Iran conflict this spring, an intelligence report circulated within the U.S. military claimed that a Chinese ship, operating in the Middle East, was transporting components related to a nuclear weapons program.

Apache helicopter on standby. Photo by Yonhap News Agency

The U.S. military began an operation to intercept the vessel. Armed personnel were reportedly ready to board the ship, and military aircraft were launched as well.

However, as the operation was about to be carried out, a review of the report revealed problems. The intelligence had been produced by an analyst from the Special Operations Command using an AI chatbot, which had incorrectly identified the cargo on board the vessel.

One source told CNN that the report was "entirely false," adding that this incident "nearly started a war." The source explained that if the U.S. military had actually carried out the operation against the Chinese vessel, it could have escalated into a military confrontation between the United States and China.

The problematic intelligence report was reportedly created after an analyst entered information about the cargo manifest of the Chinese vessel, obtained by the U.S. Pacific Special Operations Command, into an AI chatbot.

The AI analyzed both publicly available information and classified signal intelligence obtained by the U.S. government. It then concluded that the cargo was related to China's nuclear weapons program. However, this assessment itself was incorrect.

The analyst reportedly drafted the document in the standard intelligence report format used by the U.S. military, based on the findings generated by the AI, and distributed it internally.

CNN pointed out that this case demonstrates how the growing use of AI by the U.S. Department of Defense and intelligence agencies could lead to military misjudgments if AI-generated errors are not sufficiently verified by humans.

U.S. Secretary of Defense Pete Hegseth announced an "AI Acceleration Strategy" this January, under which AI is being expanded across all aspects of military work, including intelligence analysis and target selection.

However, it has been reported that the AI tools and safety standards used by various U.S. government agencies differ widely, and a unified standard for verifying AI-generated information has not yet been sufficiently established.

In particular, there is criticism that the increasing use of AI in the process of identifying military targets is problematic.

Some intelligence officials are concerned that introducing AI could actually increase the likelihood of errors during intelligence analysis. Since AI enables the faster production and distribution of information, analysts may feel increasing pressure to prioritize speed of processing.

As the application of AI expands across the military sector, this incident is likely to prompt calls for procedures to verify AI-generated information, clarify lines of responsibility, and ensure human final judgment is guaranteed.

This content was produced with the assistance of AI translation services.