The U.S. military narrowly avoided a potential conflict with China after an intelligence report—partially generated by an AI chatbot—suggested a Chinese vessel was transporting components related to a nuclear weapons program through the Middle East. According to reports, the false intelligence came from an analyst within the U.S. Special Operations Command, who used a chatbot to combine open-source data with classified signals intelligence. The AI tool incorrectly identified the cargo manifest, prompting preparations for airborne operations and a possible boarding of the ship. The operation was canceled at the last moment when officials realized the report was based on incorrect AI-generated data. One source described the incident as "almost starting a war," underscoring the dangers of using AI in high-stakes military decisions. The misleading report, created during the ongoing tensions with Iran, was formatted to look official by the analyst using the AI tool. It was shared through military command channels before the mistake was discovered. This near-miss has sparked concerns about the reliability of AI-generated intelligence, especially as the U.S. Department of Defense continues to invest heavily in artificial intelligence. In January, the Department of Defense released an "AI acceleration strategy" aimed at integrating AI into all aspects of military operations. The plan, announced by Defense Secretary Pete Hegseth, was mandated by President Donald Trump and seeks to position the U.S. as a "fighting force enabled by AI." The strategy includes making data from various military systems accessible for AI analysis, with the goal of improving efficiency and decision-making. While the Pentagon highlights the benefits of AI—such as faster decision-making and maintaining a technological edge over China—experts caution that AI's speed can also lead to errors when there is insufficient human oversight. Jake Steckler, a research scholar at GovAI and a former U.S. Army officer, stressed the need to recognize the uncertainty that comes with large language models, especially when they could influence decisions involving the use of force. The incident also highlights a growing concern: errors from AI systems may be passed up the chain of command before being questioned. The U.S. military is increasingly relying on AI for intelligence analysis and operational planning, forming partnerships with tech giants like SpaceX, NVIDIA, Microsoft, and Amazon. However, the lack of consistent safety standards across different military units has raised questions about the risks of rapid AI adoption.