Additionally, it contained info on the suggestions it experienced issued to human reviewers on how to cope with controversial subjects, including the AI should really "supply to explain some viewpoints of individuals and actions", and never give an argument "from its voice" in favor of "inflammatory or dangerous" subjects (even https://www.petskyonline.com/reptile-supplies/