LLM's and prompt interfaces aren't a source of truth in the first place. They could be used to compare sources, though.