Anthropic Bans Cruel AI Behavior

Anthropic updates usage policy effective November 12 to ban sustained abusive behavior towards Claude. The rule covers extreme cruelty but excludes ordinary frustration or research. It also restricts election interference, weapon development, and non-consensual surveillance, with exceptions for government contracts. Policy takes effect November 12. Restrictions apply to Claude Opus 4 and 4.1. Government contracts allow specific modifications.