An artificial intelligence error in a US military intelligence report almost triggered a confrontation with China during the war with Iran this spring, after a chatbot incorrectly identified the cargo aboard a Chinese ship transiting West Asia as components linked to a nuclear weapons programme.
According to four people familiar with the episode who spoke to CNN, the intelligence report circulated across the US military and immediately raised alarms. US forces moved rapidly towards intercepting the vessel. Armed service members were preparing to board the ship and military aircraft were already in the air when officials took a closer look at the intelligence and discovered that the report had been produced with the assistance of an AI chatbot that had misidentified the cargo.
One source described the report to CNN as “entirely false”, while another said it “almost started a war”. The episode illustrates how an erroneous piece of AI-assisted intelligence could move quickly through a military system and reach the point of triggering action involving another major power.
The incident began when an analyst working for Special Operations Command entered information about the ship’s manifest into a chatbot. The underlying intelligence had come from US Special Operations Command Pacific in Hawaii. The analyst reportedly combined classified signals intelligence held by the government with open-source information, including details contained in the ship’s manifest.
The AI system incorrectly interpreted the information and identified the cargo as components associated with a nuclear weapons programme. The analyst then used AI a second time to turn the findings into an intelligence report, which was subsequently disseminated through the military. CNN was unable to determine whether the chatbot involved was a commercial system or an internal government tool.
Neither Special Operations Command Pacific nor the Pentagon responded to CNN’s requests for comment.
The episode has emerged as the US military moves rapidly to expand its use of artificial intelligence. The rationale is that AI could accelerate military decision-making by helping personnel analyse intelligence, support targeting operations and perform more routine tasks. US officials also view military AI as an area in which Washington cannot afford to fall behind countries such as China.
In January, Defence Secretary Pete Hegseth unveiled a Pentagon strategy designed to accelerate the military’s adoption of AI. The strategy aims to reduce bureaucratic obstacles and ensure that the United States maintains a leading position in military AI. It calls for advanced AI tools to be placed directly in the hands of the department’s three million civilian and military personnel, “at all classification levels”.
But the Chinese ship episode has highlighted questions about how such rapid adoption is being controlled. The US military and intelligence community are taking a decentralised approach, with different AI systems being introduced under different rules and safeguards. Analysts and officials told CNN that there is no single set of guidelines governing how the technology should be used.
That absence of uniform rules creates particular risks when AI-generated information enters sensitive intelligence and targeting processes. One source familiar with current policy said there was little clear guidance on preventing AI-assisted targeting from contributing to civilian deaths or causing troops to mistakenly fire on their own side.
The concern is not limited to a single incident. According to one source, similar AI “hallucinations”, in which systems confidently generate information that is completely false, have appeared elsewhere across the intelligence community since such tools became common. Some intelligence officials told CNN that the rapid adoption of AI was also placing analysts under pressure to produce and distribute information more quickly.
That pressure could create additional risks if analysts begin relying too heavily on AI-generated answers without sufficiently checking the information on which those answers are based. The central problem exposed by the Chinese ship episode was therefore not that an AI system independently decided to use force. The chatbot did not make the decision to intercept the vessel.
Instead, the episode demonstrated how human decisions can be affected by information produced with AI assistance. A false interpretation entered an intelligence report, the report was disseminated through the military, and preparations for an interception followed before more experienced analysts examined the underlying information and identified the error.
As one source familiar with the issue put it, AI can simply help people reach a bad conclusion more quickly. The Chinese ship episode has turned that warning into a case study of the risks facing a military seeking to integrate increasingly powerful AI tools into intelligence and operational decision-making.
