Meta AI has fixed a security flaw that made it possible for users to view the prompts and their subsequent answers put forth by others to its chatbot.