Here is one particular: How to prevent ChatGPT from answering questions that are outside the scope of the provided context in the SYSTEM role message?