Anthropic announced on October 8, 2026, that its Usage Policy is scheduled to prohibit sustained and needless abusive or cruel behavior toward its models starting November 12, 2026. The company says the clause targets extreme, repeated conduct without a discernible purpose—not ordinary frustration or disagreement. Anthropic’s announcement lists the rule’s scope and exceptions.
What the rule covers—and what it excludes
Anthropic says common frustration, pushback, dark creative themes, model testing, and research are outside the new prohibition. Its description focuses on sustained, needless abuse; the rule is not a blanket ban on expressing frustration with Claude.
The clause concerns conduct toward Anthropic’s models. It does not restrict a project because it uses a particular programming language, such as C++.
A chat cutoff and an account action are different
Anthropic identifies ending persistently abusive conversations as the primary enforcement mechanism for this clause. Its broader Usage Policy separately says the Safeguards Team may warn users or throttle, limit, suspend, or terminate access when it suspects a policy violation. Those are general policy actions; the update does not make a conversation ending an automatic account penalty.
A real-time safeguard that blocks or limits an output also does not, by itself, indicate that a user violated the Usage Policy.
The policy reaches beyond Claude.ai
The Usage Policy covers people using Anthropic’s apps and Claude Code, API and developer-platform users, customers accessing Claude through cloud providers or authorized resellers, and end users of products that integrate Claude.
Anthropic’s August 15, 2025, disclosure named Claude Opus 4 and Opus 4.1 in consumer chat interfaces as models with the ability to end a rare subset of conversations. The 2026 update refers to Claude.ai and Claude Code. After a conversation ends, Anthropic’s earlier disclosure says, a user can start another chat, give feedback, or edit and retry earlier messages to create a branch. Anthropic’s 2025 explanation describes that dated feature scope.
Anthropic says it remains highly uncertain about the possible moral status of Claude and other large language models.
