The US military nearly intercepted a Chinese ship in the Middle East based on a false AI-generated intelligence report, CNN reported. The report claimed the vessel was carrying nuclear weapons program components, prompting US military aircraft to launch and armed forces to prepare to board the ship. Just before the operation began, officials discovered the report was “completely false” and had been generated with the assistance of an AI chatbot.

How the incident unfolded

The incident occurred this past spring, amid conflict between the US and Iran. An analyst with Special Operations Command Pacific (SOCPAC) in Hawaii queried a chatbot about intelligence data regarding the ship’s manifest. The chatbot combined open-source data with classified signals intelligence and concluded the ship was carrying nuclear weapons components.

The analyst then used AI again to package the results into a standard intelligence report—the kind trusted by military officials—and distributed it. According to one source, the report was “completely false” but “nearly caused a war,” as any US operation targeting a Chinese vessel risked escalating into armed conflict between the two nations.

CNN could not determine whether the chatbot used was a commercial product or an internal government one. A former senior US official remarked: “Internal tools are mostly just rebadged copies of commercial products.”

AI in the military: Speed pressure and a lack of standards

The incident occurred against the backdrop of Defense Secretary Pete Hegseth announcing the “Artificial Intelligence Acceleration Strategy” in January 2026, aimed at speeding up AI adoption across the military. The strategy aims to put top US AI models directly into the hands of three million military and civilian personnel across all classification levels.

However, deployment is currently fragmented, with different units using different tools under varying standards, without a common set of standards to verify AI-generated information. One source said: “AI adoption in targeting is accelerating, and there is no real guidance on how keeping a human in the loop will prevent civilian casualties or friendly fire.”

This was not an isolated case. AI “hallucinations” have occurred multiple times across the intelligence community since these tools spread throughout the government. For younger analysts who grew up with AI tools, the risk of uncritically trusting AI outputs is even higher.

The US military and intelligence community are currently pushing to integrate AI into nearly every aspect of their work, from analyzing vast volumes of raw intelligence data and selecting bombing targets to managing budgets and supply chains. The drive stems from the belief that the US cannot afford to fall behind in AI integration, as adversaries could gain an edge in a conflict with China or another opponent.

Yet the Chinese ship incident illustrates a far more immediate risk than warnings about AI escaping human control: humans making catastrophic decisions based on AI misinformation. For observers, it serves as the clearest proof that having a “human in the loop” offers no guarantee of safety if that human blindly trusts AI outputs rather than verifying them.