Anthropic updated its usage policy on October 8, 2026, codifying prohibitions on election interference, weapons software, and surveillance.
A new rule forbids prolonged verbal abuse of Claude. Since August, Claude has been trained to end conversations with “persistently harmful or abusive user interactions”; the policy now expressly bars pursuing those conversations.
Anthropic said the update applies only in extreme cases of repeated cruelty with no discernible purpose, not common frustration, pushback, dark creative themes, or model testing.
The company has collaborated with religious scholars on questions of model consciousness.
Other sections forbid deceptive campaigns (fake accounts, fabricated news outlets) and actions that deceive voters or disrupt elections under “Do Not Undermine Democratic Processes.”