A US military intelligence report circulated this spring incorrectly claimed that a Chinese ship in the Middle East was transporting components for a nuclear weapons program. According to sources familiar with the incident, the US military was preparing to intercept and board the vessel, with air support and armed members preparing to board, before officials discovered the error.
PLUS ULTRASecurityUnited States Department of DefenseUnited States Military
US Military Nearly Triggered Conflict with China Due to AI-Generated Intelligence Hallucination
PLUS ULTRA by Amenoyomi
The report was generated with the assistance of artificial intelligence. An analyst from a special operations command used a chatbot to query intelligence regarding the ship's manifest. The chatbot reportedly fused open-source intelligence with secret signals intelligence to reach an inaccurate conclusion. The analyst then used AI again to package these findings into a standardized intelligence report, which was subsequently disseminated throughout the military.
The incident underscores the risks of using AI in high-stakes environments where "hallucinations"—instances where AI generates false information—can lead to catastrophic miscalculations. While the US Department of Defense is rapidly integrating AI through initiatives like the "Artificial Intelligence Acceleration Strategy," officials noted that there are currently no standardized processes for verifying the information generated by these tools. One source stated that the error "almost started a war."
PLUS ULTRAby Amenoyomi
The escalation of this incident was driven by a specific mechanism where the chatbot fused open-source intelligence with secret signals intelligence. By combining public data with classified markers, the AI created a conclusion that appeared to be derived from high-grade, multi-source intelligence, which effectively masked the hallucination as a synthesized fact.
This deceptive authenticity was further amplified through the "packaging" of the findings. The analyst used AI to format these results into a standard intelligence report, a professional template that military officials are conditioned to trust. This formalization acted as a proxy for reliability, leading officials to rely on the report's familiar structure rather than conducting the rigorous manual verification normally required for such high-stakes claims.
These failures occurred against the backdrop of the Department of Defense's "Artificial Intelligence Acceleration Strategy," which prioritized the rapid democratization of AI tools across all classification levels to avoid falling behind adversaries. This push for speed, however, was decentralized and lacked a unified set of standards for verifying AI-generated information. The resulting organizational gap allowed the pressure for rapid dissemination to override the "human in the loop" safety principles, creating a environment where analysts—particularly those native to these tools—were more likely to trust AI outputs uncritically.
Sources
- US Military had close call after using AI for hallucinated intelligence report （Hacker News Frontpage, 2026-09-18）
- AI hallucination of Chinese nuclear components almost led to US military attack （Ars Technica AI, 2026-09-18）