In the spring of 2026, US military forces nearly intercepted a Chinese vessel in the Middle East due to an erroneous intelligence report generated by AI. According to CNN, the document falsely claimed the ship was transporting components for a nuclear weapons program. Military personnel were prepared to board the vessel, and aircraft were already deployed before analysts identified the error.
The incident occurred after an analyst uploaded shipping manifest data into an AI chatbot. The system cross-referenced open-source information with classified signals intelligence, resulting in an incorrect assessment. The analyst then used the AI to format these findings into a standard report. It remains unclear whether the tool used was a commercial model or a government-internal system.
Sources indicate that the use of AI tools within US military and intelligence agencies remains decentralized, with no unified standard for verifying generated information. Similar AI hallucinations have been reported previously. While the military is increasingly using neural networks for intelligence analysis and target selection, there are currently no clear regulations regarding human oversight to prevent erroneous actions.
In June, Bloomberg reported that the Pentagon had secretly approved an updated doctrine for the use of AI in battlefield target selection. Despite the risks associated with AI errors, the integration of these technologies into military operations continues to expand across various departments.
