You’re asking ChatGPT for advice to stop drone attacks? Does that mean people die if it hallucinates a wrong answer and that isn’t caught?
Military technologies are validated before deployed. Nobody can die from a hallucination.
But if I want to understand, say, how a particular Russian drone works, ChatGPT can help me piece together information from English, Russian, and Ukrainian-language sources.
But sometimes ChatGPT's safety filter thinks I want to use the Russian drone instead of stopping it, in which case it doesn't want to help.