Please explain why the mechanism of the LLM generating output precludes it from being able to be fooled without using use tautologies or reducing to substrate for explanation.
LLMs don't hold beliefs (neither do mechanistic processes), and they aren't a someone.
You can widen out the definition words but that generally makes language weaker - interestingly, semantic drift is a big issue for LLM's.