What did Anthropic say about abusing Claude AI? New policy bans ‘cruel’ behavior

Anthropic has updated its usage policy to prohibit users from engaging in "sustained and needless abusive or cruel behavior" towards its Claude AI models. The new rules will take effect on November 12, 2026. The policy bans users from repeatedly subjecting Claude to "needless cruelty or 'abusive' behavior," though it does not define abusive behavior explicitly.

Users can still challenge Claude's answers, express dissatisfaction, test its capabilities, or explore "dark themes" for creative purposes. Anthropic also clarified its requirements for high-risk use cases in healthcare, finance, and autonomous physical actions. The revised policy expands restrictions on deceptive political campaigns, election interference, weapons development, and unauthorized surveillance, consolidating these rules under a new section called "Do Not Engage in Deceptive Campaigns or Artificial Activity."

The update is an extension of Anthropic's August 2025 decision allowing Claude to end conversations with persistent harmful or abusive interactions, which was part of the company's research into potential AI welfare.