[selcuka]

Sure, but LLMs are inherently lossy. There is no guaranteed way for the second AI to extract the original prompt from the text.