AI-hallucinated intel report nearly led US troops to board Chinese ship

According to CNN, an analyst used a chatbot to interpret intelligence on a Chinese vessel's cargo, and the system produced false claims of nuclear weapons aboard. US troops prepared to board the ship with military aircraft support before the error was caught just before the operation. Sources called the assessment "entirely false" and said it "almost started a war."

• Chatbot combined public material with classified signals and produced a false cargo assessment
• Armed personnel prepared to board the vessel while military aircraft supported the operation
• The error was found before the operation; the Pentagon and Pacific command declined to comment
• US military uses GenAI.mil: over 1.3 million personnel and tens of millions of prompts