The US military came close to intercepting a Chinese vessel in the Middle East this spring after an AI-assisted intelligence report falsely identified its cargo as components linked to a nuclear program, CNN reported, citing sources familiar with the incident.

US aircraft were already in the air and preparations for the interception were underway when military personnel decided to conduct a final review of the intelligence. That check revealed that the analytical chatbot had misidentified the cargo, prompting the operation to be called off.

One CNN source described the report as entirely false and said the incident nearly triggered a war between the United States and China. CNN was unable to determine what the Chinese vessel was actually carrying.

The error originated during an analyst’s review of intelligence provided by US Special Operations Command Pacific, according to CNN. The analyst used a chatbot to process the material, combining information from open sources with classified signals intelligence. The system then produced an incorrect assessment of the cargo aboard the Chinese ship.

AI was used again to prepare the final intelligence report. The document was subsequently distributed in a format that US military personnel would typically consider reliable.

CNN reported that the flawed assessment could have had far-reaching consequences. An American operation against a Chinese vessel carried the risk of escalating the incident into a direct confrontation between the United States and China.

The incident comes as the US military and intelligence community increasingly turn to artificial intelligence for intelligence analysis and target selection. Different units use different AI systems and security standards, while no uniform process is currently in place to verify AI-generated intelligence.

A CNN source said the use of AI in target selection is expanding rapidly, even as clear guidelines designed to prevent errors have yet to be established.