Anthropic’s Claude Cruelty Ban Takes Effect November 12

Anthropic announced October 8 that a revised Claude usage policy will take effect November 12, prohibiting users from repeatedly subjecting its models to purposeless cruelty. The narrow rule does not cover ordinary frustration or criticism, dark creative work, or legitimate testing and research. It formalizes Claude’s existing ability to end persistently abusive conversations, while Anthropic has not specified whether other penalties, such as account bans, will apply. The broader update targets deceptive political and commercial campaigns, voter misinformation and election disruption, and expands restrictions on weapons-related uses, surveillance, and other high-risk applications. The model-welfare provision has drawn attention amid debate over whether AI could experience suffering, as well as criticism that tech companies should prioritize harms to people.