Can a technology called RAG keep AI models from making stuff up?

I think I saw some research recently that showed a few examples of how RAG actually increases hallucination in some cases.