An AI chatbot nearly triggered a conflict between China and the USA.

A CNN revelation has brought to light an almost unbelievable episode: an artificial intelligence chatbot nearly triggered a military conflict between the United States and China. According to the reconstruction, an American intelligence officer asked the AI to draft a report on naval traffic. In response, the bot generated completely false information, accusing a Chinese ship of transporting material useful for the manufacture of atomic weapons to Iran.

This scenario, if taken as true, would have required immediate military intervention by American forces against the Chinese vessel, with the concrete risk of triggering a large-scale war. The incident was stopped only minutes before the so-called 'red button' could be pressed, when it emerged that all the information provided by the chatbot was flawed and the vessel was not carrying any military materials.

The episode fits into the global debate on the risks associated with the use of artificial intelligence, particularly so-called 'chatbots' or conversational assistants. These systems, based on advanced language models, are designed to generate text in response to human prompts, but they can produce inaccurate or entirely fabricated information, a phenomenon known as AI 'hallucination'. The technology originated from research in natural language processing and has seen rapid dissemination in recent years, finding application in various sectors, including national security.

The underlying problem is that these systems, while extremely sophisticated, do not possess a real understanding of the world and can confabulate plausible but false answers. When used in high-risk contexts such as military intelligence, where decisions are based on critical information and tight timelines, a single error can have catastrophic consequences. The incident described by CNN shows how reliance on automated tools, without adequate human oversight and cross-validation, can bring nations to the brink of international conflict based on completely false premises.