US Military Nearly Boarded Chinese Ship After AI Misidentified Cargo

Armed forces prepared to intercept vessel based on false AI analysis

The US military came dangerously close to boarding a Chinese ship this spring after an artificial intelligence system incorrectly identified its cargo as nuclear weapons components, according to four sources familiar with the incident.

Armed military personnel were preparing to intercept the vessel in the Middle East during ongoing conflict with Iran, and military aircraft were already airborne before officials discovered the intelligence report was fundamentally flawed. The report, which one source described as "entirely false," was generated with AI assistance by a special operations analyst.

Officials only caught the error moments before the planned operation. "It almost started a war," one source told CNN, which first reported the details. Any US action against a Chinese vessel could have escalated into armed conflict between the two powers.

How the AI error occurred

The incident began when an analyst at US Special Operations Command Pacific in Hawaii queried a chatbot about intelligence on the ship's manifest. The system—either a commercial product or government tool—combined open-source information with classified signals intelligence and reached an incorrect conclusion about the cargo.

The analyst then used AI again to format the findings into a standard intelligence report format that military officials typically trust, then distributed it through normal channels. Sources could not confirm what the misidentified cargo actually was.

A former senior US official familiar with military AI systems told CNN that "the internal tools are mostly just copies of the commercial stuff wearing lipstick," suggesting government systems may not be substantially more reliable than commercial alternatives.

Decentralized AI adoption across military

The episode highlights risks in the military's rapid, decentralized AI deployment. Different military branches and intelligence agencies use different tools under varying safety standards, with no unified verification process for AI-generated information.

Defense Secretary Pete Hegseth released an "Artificial Intelligence Acceleration Strategy" in January 2026 aimed at putting AI "directly in the hands of our three million civilian and military personnel, at all classification levels." The strategy emphasizes speed and broad experimentation across the department.

Sources say this type of AI "hallucination" has occurred multiple times across the intelligence community. The technology creates pressure on analysts—particularly younger ones more comfortable with AI tools—to produce intelligence faster, potentially without adequate verification.

"AI allows you to get to a bad idea faster," one source said.

Why it matters

This incident represents a concrete example of AI risk that differs from theoretical concerns about artificial general intelligence. Rather than AI systems acting autonomously, the danger lies in humans making catastrophic decisions based on flawed AI outputs—especially in targeting and military operations where mistakes can trigger international conflict. With the military rapidly expanding AI use in battlefield decision-making to compete with adversaries like China, the lack of standardized verification protocols across different military AI systems creates immediate operational risks.

US Special Operations Command Pacific and the Pentagon did not respond to requests for comment on the incident, according to CNN's reporting.