We should be able to automatically detect if some sort of phishing was being attempted in a conversation