is there a way for the AI to understand if it said misinformation?