Furthermore, it contained info on the suggestions it had issued to human reviewers on how to manage controversial subjects, such as which the AI must "offer to describe some viewpoints of people and actions", and not supply an argument "from its voice" in favor of "inflammatory or hazardous" topics (although https://gunnernvbgl.thekatyblog.com/28999472/the-smart-trick-of-chat-gpt-login-that-no-one-is-discussing