The United States military came closer to a direct confrontation with China than nearly anyone realized this spring — and according to a CNN exclusive published Friday, an AI hallucination was the reason why.
The report details how an intelligence document generated with the help of a chatbot set off a chain of events that had armed US personnel preparing to board a Chinese ship in the Middle East, on the belief that the vessel was transporting components of a nuclear weapons program. Military aircraft were already in the air, according to sources familiar with the operation. For more context on this story, see our ongoing AI industry coverage.
The mission was scrapped only just before the planned interception, when officials dug deeper into the report and discovered that its central claim — the one that had mobilized the US military — was wrong.
One source familiar with the episode described the intelligence report to CNN as "entirely false." It also, the source said, "almost started a war." Any US operation against a Chinese vessel could have spiraled into an armed conflict between the two nations.
How a Chatbot's Claim Became a Military Operation
The episode began this spring, amid the war with Iran, when an intelligence report began circulating across the US military. The document claimed a Chinese ship in the Middle East was carrying nuclear weapons program components, and it immediately set off alarm bells.
According to CNN, the report was put together by an analyst at US special operations command who had queried a chatbot about intelligence reporting on the ship's manifest — reporting that originated with US Special Operations Command Pacific, based in Hawaii. It remains unclear whether the chatbot was a commercially available product or a US government tool. A former senior US official familiar with the AI systems used by military and intelligence analysts offered CNN a blunt assessment of the government's internal offerings: "The internal tools are mostly just copies of the commercial stuff wearing lipstick."
The chatbot fused open-source intelligence with secret signals intelligence held by the government and reached its fateful conclusion about the ship's cargo. The analyst then used AI a second time to package the findings into a standard intelligence report — the kind military officials are accustomed to trusting — and disseminated it.
Four sources familiar with the episode spoke to CNN. According to two of them, armed members of the US military were preparing to board the ship when the reporting fell apart. CNN was unable to learn what the misidentified cargo actually was. US Special Operations Command Pacific and the Pentagon did not respond to CNN's request for comment.
An Acceleration Strategy Meets a Reality Check
The near-miss lands at a moment when the Pentagon is pushing to weave AI into nearly every facet of its work — analyzing huge volumes of raw intelligence, selecting targets for strikes, and handling more mundane tasks like budgeting, logistics and supply chains.
In January, Defense Secretary Pete Hegseth released the department's "Artificial Intelligence Acceleration Strategy," a bid to speed up military AI adoption.
"We will unleash experimentation, eliminate bureaucratic barriers, focus our investments and demonstrate the execution approach needed to ensure we lead in military AI," Hegseth said in a speech announcing the strategy.
The accompanying memo goes further, ordering the department to make AI available through several programs with the aim of "democratizing AI experimentation and transformation across the Department by putting America's world-leading AI models directly in the hands of our three million civilian and military personnel, at all classification levels."
But the effort is decentralized, multiple US officials told CNN, with different parts of the government using different tools under different orders and safety standards. There is no single set of standards for verifying information generated by these tools, and the reliability of the various systems varies widely.
Targeting Is Where the Risk Lives
Analysts have long feared that AI could contribute to a catastrophic miscalculation if nation-states act on poor or corrupted data. The Chinese ship episode is a documented case of exactly that failure mode forming in real time — and of the safeguards failing to catch it until the last moment.
"AI in targeting is definitely something that is ramping up, and there is no real guidance for how having a human in the loop will prevent civilian casualties or fratricide," another source familiar with the military's current policies told CNN.
According to CNN's reporting, the kind of hallucination that produced the false claim has not been an isolated incident across the intelligence community since AI tools began proliferating through government. Several sources also pointed to a cultural risk: younger analysts, described as natives of these tools, are more likely to trust them uncritically, and the pressure to produce intelligence faster leaves more room for error.
"AI allows you to get to a bad idea faster," one source said.
A More Immediate Risk Than Superintelligence
For weeks, Washington has been consumed by a debate over long-term AI risk, following a series of dire public warnings from Silicon Valley engineers and tech executives about systems potentially breaking free of human constraints. The CNN report underscores a different and more immediate danger: human beings making disastrous decisions based on inaccurate or misleading AI output — right now, in the middle of real geopolitical tensions.
No shots were fired this spring, and the boarding never happened. But the episode stands as the clearest example yet that machine-generated errors, once laundered into official reporting, can travel through the world's most powerful military apparatus all the way to the edge of armed conflict between nuclear powers — with no standing verification standard in place to stop the next one.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
