How An AI Chatbot Seemingly 'Almost Started A War'

Depending on your point of view, some older fictional movies were less works of fantasy and more warnings about the future. For instance, the movie "WarGames" is the story about a hacker who almost starts World War III after he unwittingly unleashes an AI that feeds the U.S. military false information. That movie inspired Ronald Reagan's cybersecurity policies, but it also almost predicted a recent event beat for beat. Sort of.

On September 18, CNN reported that we had narrowly escaped a veritable doomsday scenario. According to the story, the U.S. military circulated an intelligence report that a Chinese ship in the Middle East was secretly shipping nuclear weapon components. The U.S. military sprung into action and was ready to board the vessel. Airplanes were already en route to intercept, but then word came that the mission should be scrubbed because, after officials dug a little deeper, they discovered that the entire mission was based on faulty information. Specifically, the claims that the ship was carrying nuclear material were the product of a hallucinating AI chatbot. We don't know what the ship was carrying, but it certainly wasn't anything that could be transformed into a nuclear bomb.

One of CNN's sources close to the event said that had U.S. forces boarded the vessel, it could have "started a war." According to CNN, military analysts have been terrified at the notion of AI providing "poor or corrupted data" that could result in a "catastrophic miscalculation." This close call could be viewed as proof that they were right all along. It's true what the AI in "WarGames" said regarding thermonuclear war (and tic-tac-toe) at the end of the film: "The only winning move is not to play."

In 2025, we reported on an op-ed that stated AI hallucinations would only get worse from there on out. If our brush with World War III over a chatbot misidentifying Chinese cargo is any indication, the article sugarcoated the potential damage. But that still raises the question of how an AI model brought us to the brink of another unnecessary armed conflict.

First and foremost, officials throughout the U.S. military and intelligence sectors are trying to implement AI into every level, including the examination and evaluation of data. In this situation, an unnamed analyst used a chatbot to combine "open-source intelligence with secret signals intelligence in government holdings" and turn it into an intelligence report — one that said the Chinese ship in question was ferrying nuclear material.

Many people believe that AI is becoming essential for job security (and general job duties), the U.S. government included. Officials believe that, due to increasing amounts of data, the military needs AI to decide what targets are worth pursuing — and striking. However, AI adoption is "decentralized" and lacks a common standard for verifying AI-generated information. Different parts of the government use different tools, which makes their results unreliable. It doesn't help matters that the chatbots government analysts use are "just copies of the commercial stuff wearing lipstick," according to an unnamed former senior U.S. official. To make matters worse, the hallucination that almost resulted in the prelude to World War III was not an isolated incident; CNN's sources claim AI tools have produced a series of false conclusions ever since they were implemented into government analyst toolkits. The next time, we might not be so lucky.