Anthropic announced on October 8, 2026, that its revised Claude Usage Policy is scheduled to take effect November 12. The new provision prohibits “sustained and needless abusive or cruel behavior” toward Anthropic’s models, which the company says means extreme cases where users repeatedly act cruelly without a discernible purpose. Anthropic’s policy update
What the rule covers—and what it excludes
Anthropic explicitly excludes common user frustration, pushback, dark creative themes, model testing, and research from the new provision. It focuses on repeated cruelty in extreme cases, rather than ordinary friction in a conversation.
The provision addresses behavior toward the models; it does not single out a programming language or project type. Anthropic also lists model testing and research among the exclusions.
What happens when Claude ends a chat
Anthropic says Claude’s ability to end a conversation remains the primary enforcement mechanism for the conduct covered by the policy. The feature predates the revision: in a post published August 15, 2025, Anthropic described it for Claude Opus 4 and Claude Opus 4.1 in consumer chat interfaces. The company said it should be a last resort after repeated attempts to redirect the conversation fail, with an exception when a user asks Claude to end the chat. Anthropic also said Claude should not use the feature when a user may be at imminent risk of harming themselves or someone else. Anthropic’s 2025 feature description
In that 2025 description, ending a chat prevented further messages in that thread but did not affect other conversations on the account. Users could start a new chat or edit and retry earlier messages to branch from the ended conversation.
What the policy says about Claude’s moral status
A conduct rule is not a finding that Claude is conscious or suffers. In its 2025 feature note, Anthropic said it remained highly uncertain about Claude’s potential moral status.
