AI-Generated Intelligence Nearly Triggered a U.S. Military Boarding of a Chinese Ship

The United States narrowly avoided boarding a Chinese ship after an intelligence report, prepared with help from AI tools, incorrectly claimed that the vessel was transporting components linked to a nuclear weapons program in the Middle East.

According to a CNN report, four sources familiar with the episode said the false information was submitted by an analyst with U.S. Special Operations Command. As U.S. forces prepared to intercept and board the ship with air support, authorities discovered that the chatbot used to prepare the report had inaccurately identified the supplies listed in the vessel’s manifest.

One source told CNN that the AI-related failure “almost started a war.”

How an AI intelligence error nearly escalated

CNN reports that analysts used chatbots to analyze intelligence about the Chinese ship’s manifest. The chatbot reportedly “fused open-source information with secret government-held signals information,” producing an intelligence report that nearly set off a potentially disastrous chain of events.

The incident is one of the most serious examples yet of how AI hallucinations can undermine professional reports and high-stakes decisions. AI systems can generate confident but false information when they lack sufficient context or misinterpret their training data.

Since “hallucination” became Cambridge Dictionary’s word of the year in 2023, similar problems have been documented involving nonfiction writers, journalists, academic researchers, judges, doctors, police departments, corporate call centers, and other organizations. Some researchers suggest that completely preventing large language model hallucinations may not be possible.

AI use in military intelligence raises new risks

The U.S. military is expected to account for these limitations when using AI to analyze intelligence. However, the Department of Defense rolled out its AI Acceleration Strategy in January, with the goal of making “all the right data available across AI-enabled collaborative IT systems, including mission systems across all services and components.”

The near-miss highlights the risks of relying on AI-generated analysis in situations where an incorrect conclusion could trigger military action, escalate tensions with China, or create an international crisis.