You should pretty much assume everything that a chatbot says could be false to a much higher degree than human written content, making it effectively useless for your stated purpose.