I think I couldn’t explain it better. My question is that, my databse/documents will not contain any sensitive/blacklisted question, but due to the nature of smart AI, it tries to answer some basic genral question which is out of the context of the database. How to prevent it?