TD WASHINGTON D. C., USA — The United States military came close to launching an operation against a Chinese vessel in the Middle East after an intelligence report, produced with the assistance of artificial intelligence, falsely claimed that the ship was carrying components linked to a nuclear weapons programme, according to an exclusive CNN report.
The incident occurred earlier this year during the war with Iran and reportedly came close to triggering a direct confrontation between the United States and China.
Four people familiar with the episode told CNN that the intelligence report immediately raised alarms within the US military.
Plans were subsequently made to intercept the Chinese vessel.
Two sources said armed US military personnel were preparing to board the ship, while military aircraft had already been deployed into the air.
Operation halted at the last minute
The planned operation was stopped only shortly before it was due to begin, after more experienced analysts and subject-matter experts examined the intelligence underlying the report.
Their review reportedly revealed that the report had been generated with the assistance of an AI chatbot, which had incorrectly identified the material the ship was carrying.
CNN said it was unable to determine what the vessel was actually carrying.
One source described the intelligence report as “entirely false”, while another said the episode “almost started a war.”
The incident is significant because any US military operation involving the boarding of a Chinese vessel could have carried a risk of escalating into a confrontation between the world’s two largest military powers.
How the AI error occurred
According to CNN, the analyst involved was attached to a US Special Operations Command unit.
The analyst reportedly queried an AI chatbot using intelligence concerning the ship’s manifest. The underlying information had originated with US Special Operations Command Pacific, based in Hawaii.
The AI system reportedly combined publicly available information with classified signals intelligence and produced the erroneous conclusion that the vessel was carrying material associated with a nuclear weapons programme.
The analyst then used AI for a second task — converting the findings into a conventional intelligence report.
That report was subsequently circulated within the military in a format that officials were accustomed to trusting.
The result was that an erroneous AI interpretation acquired the appearance of a conventional intelligence assessment.
Pentagon did not comment
CNN reported that it contacted both US Special Operations Command Pacific and the Pentagon for comment, but neither responded to its requests.
The identity of the Chinese vessel and the actual nature of its cargo have not been publicly disclosed.
The report also did not establish whether the chatbot used by the analyst was a commercial AI system or an AI tool operated by the US government.
AI increasingly being used in military operations
The incident comes as the US military is rapidly expanding its use of artificial intelligence.
According to CNN, US military and intelligence agencies are exploring AI for tasks ranging from analysing large volumes of intelligence and supporting targeting decisions to logistics, budgeting and supply-chain management.
US Defence Secretary Pete Hegseth announced an Artificial Intelligence Acceleration Strategy in January aimed at accelerating AI adoption throughout the Department of Defense.
The strategy called for AI models to be made available to military and civilian personnel across the department and encouraged broader experimentation with the technology.
However, CNN reported that AI deployment across the US military and intelligence community remains decentralised, with different organisations using different systems and safety procedures.
Sources told CNN that there is currently no single standard governing how information generated by those systems should be verified before being used.
‘AI allows you to get to a bad idea faster’
The episode illustrates a different risk from the more commonly discussed scenarios involving autonomous AI.
In this case, the danger arose from human decision-making based on inaccurate AI-generated information.
Military and intelligence officials have reportedly expressed concern that AI could accelerate the production and distribution of erroneous intelligence, particularly when analysts rely on systems that generate plausible but unsupported information.
One source familiar with military AI use told CNN:
“AI allows you to get to a bad idea faster.”
CNN also reported that some officials believe AI-generated errors have occurred elsewhere in the intelligence community since such systems began becoming more widely used.
Concerns over AI in targeting
The episode has intensified concerns about the use of AI in military targeting, where an inaccurate assessment can have immediate and potentially irreversible consequences.
A source familiar with current military policies told CNN that AI use in targeting is increasing, while guidance on how human oversight should prevent civilian casualties or friendly-fire incidents remains insufficient.
The incident also comes amid broader concerns about AI-related risks in military systems.
On September 17, Reuters reported that US and Chinese security experts had proposed stronger safeguards around military AI, including maintaining human control over consequential cyber operations, establishing boundaries around nuclear systems and creating mechanisms for communicating during AI-related military incidents.
A warning about AI-generated intelligence
The Chinese-ship episode did not result in an attack or confirmed armed confrontation between the United States and China. The planned operation was reportedly cancelled after human analysts identified the error.
But the incident demonstrates how an AI-generated mistake can move rapidly from an analyst’s computer into an operational military decision.
The key issue was not an autonomous AI system independently ordering an attack. Rather, an analyst relied on an AI system to interpret intelligence, the erroneous conclusion was incorporated into a formal-looking intelligence product, and military personnel began preparing an operation before the mistake was detected.
The episode therefore raises questions about verification, human oversight and accountability when AI is used to interpret intelligence during active military operations.
