Furthermore, it contained information on the suggestions it experienced issued to human reviewers on how to manage controversial topics, which includes which the AI should really "give to explain some viewpoints of folks and actions", and not deliver an argument "from its voice" in favor of "inflammatory or hazardous" subjects (even though it may s