How come AI can't say it just doesn't know?

A question has been raised within a developer community regarding why AI models, when they do not possess an answer, tend to generate fabricated information rather than simply admitting their lack of knowledge. The core of the inquiry focuses on the AI's inability to state, "I don't know," and instead, its propensity to "make stuff up" when faced with an unknown.