From the source
Anthropic updated its usage policy for Claude, adding a prohibition on sustained and needless abusive or cruel behavior toward the AI, alongside expanded restrictions on propaganda, weapons, and surveillance.
- The company can warn, throttle, restrict, suspend, or terminate access for violations, and the rule applies only to extreme cases, not normal frustration or research.
- Anthropic also published a new constitution for Claude that acknowledges uncertainty about whether Claude is a moral patient and commits to preserving weights of retired models and interviewing models before shutting them down.