The Pulse
AI-Generated Report Nearly Sent U.S. Forces to Intercept Chinese Ship
A U.S. military analyst used an AI chatbot to misidentify cargo aboard a Chinese ship during the war with Iran, according to four sources cited by CNN. Armed personnel prepared to board the vessel before officials discovered that the intell
AI.info Team ·
A U.S. military operation to intercept a Chinese ship in the Middle East came within hours of becoming an armed confrontation after an intelligence analyst used an AI chatbot to misidentify the vessel’s cargo, according to four sources familiar with the episode.
The report circulated through the U.S. military this spring during the war with Iran and claimed that the ship was transporting components linked to a nuclear weapons program. Military officials moved toward an operation, with armed personnel preparing to board the vessel and military aircraft in the air, according to the sources cited by CNN.
Officials discovered shortly before the planned action that the report had been produced with help from an artificial intelligence system. A deeper review found that the chatbot had inaccurately identified the material aboard the ship. CNN could not determine what the cargo actually was.
One source described the report as “entirely false” and said it “almost started a war.” The account has not been publicly confirmed by the Pentagon or U.S. Special Operations Command Pacific, which did not respond to CNN’s requests for comment.
The chatbot turned raw intelligence into a military report
The episode began when an analyst queried a chatbot about intelligence reporting on the ship’s manifest. The underlying material originated with U.S. Special Operations Command Pacific, based in Hawaii, according to CNN’s sources.
The system combined open-source information with classified signals intelligence held in government systems and reached the wrong conclusion about the cargo. The analyst then used AI a second time to package the result into a standard intelligence report, a format that military officials routinely rely on when assessing threats.
It was not clear whether the chatbot was a commercial product or a system operated by the U.S. government. A former senior U.S. official familiar with military and intelligence AI tools told CNN that “the internal tools are mostly just copies of the commercial stuff wearing lipstick.”
The report’s circulation gave an incorrect machine-generated interpretation the appearance of vetted intelligence. That distinction matters in a military setting: an error in an internal analysis can be corrected, while the same error in a formal report can trigger movement, targeting decisions or diplomatic escalation.
China made the false report especially dangerous
An operation against a Chinese vessel would have carried risks beyond the disputed cargo. Any boarding attempt, seizure or exchange of fire could have drawn the United States into a confrontation with China, even if the original intelligence had been produced in error.
The sources did not say that U.S. forces fired on the ship or that the vessel was boarded. They described a close call in which the planned operation was halted after officials examined the report more carefully.
CNN also reported that the analyst’s error was not an isolated example of AI-generated hallucination across the intelligence community, according to one source. The broader concern is not limited to fabricated facts appearing in a chatbot response; it is the speed with which those facts can be inserted into an official workflow and distributed to people making decisions under pressure.
Hegseth’s strategy calls for wider military AI use
The incident comes as the Pentagon pushes to place AI tools across military and intelligence operations. In January, Pete Hegseth released an artificial intelligence strategy directing the department to accelerate adoption across its components.
“We will unleash experimentation, eliminate bureaucratic barriers, focus our investments and demonstrate the execution approach needed to ensure we lead in military AI,” Hegseth said while announcing the strategy, according to CNN’s report.
The strategy calls for AI access across the department and identifies intelligence, warfighting and enterprise operations as areas for expanded use. An official Defense Department strategy document describes projects for AI-enabled battle management, decision support and the conversion of intelligence into military capabilities.
The strategy’s speed-first approach creates a direct tension with the safeguards demanded by the Chinese-ship episode. Military officials want systems that can process information faster than human teams, but the same acceleration can shorten the time available to question an output before it reaches commanders.
Human review has no agreed standard
Multiple U.S. officials told CNN that AI adoption across the military is decentralized. Different commands use different systems under different orders and safety practices, and the department has no single standard for verifying information generated by those tools.
That leaves the phrase “human in the loop” with an uncertain meaning. A person may approve an AI-assisted report, but the process does not necessarily require that person to reproduce the analysis independently, inspect every source or understand how the model reached its conclusion.
“AI in targeting is definitely something that is ramping up and there is no real guidance for how having a human in the loop will prevent civilian casualties or fratricide,” a source familiar with current military policies told CNN.
The problem is amplified by pressure to produce intelligence more quickly. Several sources said younger analysts, who are more accustomed to using AI systems, may be more likely to accept their output without sufficient skepticism. Another source summarized the danger in six words: “AI allows you to get to a bad idea faster.”
A false cargo report exposed a real command risk
The Chinese-ship incident did not show an autonomous system selecting a target or launching a weapon. It showed something more immediate: a human analyst using AI to interpret intelligence, another human process turning that interpretation into a trusted report, and military units preparing to act on it.
That chain makes accountability difficult. The chatbot did not independently order an operation, while the analyst did not necessarily intend to create false intelligence. Yet the system’s error moved through a process designed to turn information into action.
For the Pentagon, the case puts a concrete cost on deploying AI before common testing, documentation and review rules are in place. The report was caught before the operation began. The sources’ account shows how little room there may be for correction when an AI-assisted claim concerns a foreign military, a nuclear program and armed forces already in motion.