Anthropic's updated usage policy prohibits sustained, needless, and extreme cases of repeated cruelty or abuse toward Claude AI models with no discernible purpose, excluding common user frustration, pushback, dark creative themes, or research