A U.S. special operations analyst reportedly used AI to write an intelligence report so wrong it nearly triggered a diplomatic crisis with China.
According to CNN, cited secondhand by PCMag, an analyst at U.S. Special Operations Command compiled the report with help from an AI tool. The report contained misleading information involving China, and CNN describes it as having almost started a war. PCMag's writeup does not include a timeframe for when this happened, how the error was caught, or what happened afterward. No official statement, document, or named source appears in the available reporting.
Military and intelligence agencies have been quietly experimenting with generative AI to speed up analysis, and this is the kind of incident critics have warned about: a model producing plausible-sounding but wrong output that a human then signs off on. The stakes are different than a chatbot getting a fact wrong in a customer-service window - a flawed intelligence assessment can shape decisions with real geopolitical consequences.
As reported, the story rests on a single CNN account relayed through PCMag, with no corroborating details on timing or resolution - worth watching for a fuller account before treating 'almost started a war' as more than a dramatic paraphrase.