Anthropic updated its Usage Policy on October 8, 2026, prohibiting sustained, needlessly abusive or cruel behavior toward AI models including Claude, applying only to extreme cases of abuse while excluding frustration, pushback, dark themes, testing, and research