Anthropic updated its usage policy to ban sustained abusive behavior toward Claude. The rule applies only to extreme cases with no clear purpose. Common frustration and model testing remain allowed. Claude can still end abusive conversations on its own. The policy also tightens rules on election interference and weapons development. It bans using Claude for surveillance or deceptive influence campaigns. Personalized political targeting is now allowed if not deceptive. Weapons bans now explicitly cover software and components. Physical equipment connected to Claude must have human oversight. The changes take effect on 12 November 2026.
What changed
Claude could end abusive conversations but had no explicit user policy against cruelty.
What it unlocks
Anthropic can now block users for sustained abusive behavior toward its models.
- policy takes effect 12 November 2026
Sources