False intelligence report nearly triggered a military war

@Mark Ollig

WASHINGTON — An undisclosed artificial intelligence chatbot tool used by U.S. Special Operations Command Pacific generated a false intelligence report that incorrectly alleged a Chinese commercial vessel was transporting nuclear components.

The AI-generated report nearly triggered a military conflict, but senior officials aborted a forced boarding of the ship after a last-minute data review.

Anonymous sources told CNN that an AI chatbot fabricated a false intelligence report, nearly causing a military conflict between the United States and China.

The incident occurred in May during the U.S. conflict with Iran, when a military intelligence analyst with U.S. Special Operations Command Pacific used an internal AI tool to process the cargo manifest of a Chinese commercial vessel.

The chatbot, designed to merge classified signals intelligence with open-source data, falsely claimed that the ship was transporting components for a nuclear weapons program.

The chatbot has not been identified. Sources said it was built to synthesize disparate data streams, combining open-source data with classified signals intelligence.

The AI tool formatted its fabricated findings into a standard intelligence report.

Because the document mirrored authentic military intelligence products, it circulated rapidly through official channels absent immediate skepticism.

Believing they were responding to an active nuclear proliferation threat in a combat zone, the U.S. military scrambled warplanes and positioned heavily armed personnel to board and seize the Chinese vessel.

Forcibly boarding a sovereign Chinese ship in international waters carried immense geopolitical risk and could have been interpreted as an act of war.

The operation was aborted just minutes before the assault team made contact. A last-minute review of the raw data by senior military officials revealed that no human expert had ever verified the core allegation and that it was entirely the chatbot's fabrication.

Command units ordered an immediate stand-down, and the ship's cargo remains undetermined.

Defense officials familiar with the event warned that the automated error nearly started a war, fueling intense internal debate at the Pentagon over the military's rapid adoption of generative AI.

Current and former officials stated that many military AI tools are essentially repurposed versions of unreliable commercial software.

A former senior U.S. official cautioned that many internal applications used by military analysts are essentially commercial software with minor modifications, relying on large language model designs that suffer from the same unreliability and hallucination flaws as public systems.

They also warned that younger analysts face pressure to distribute intelligence quickly, often leading them to trust automated outputs without adequate verification.

AI-generated illustration created with ChatGPT.
Based on CNN reporting published by CP24 on Sept. 18, 2026.
This is an illustrative reconstruction, not a photograph of the incident.