Anthropic bars sustained, needless abuse of Claude from Nov 12; criticising answers is allowed

People put all kinds of questions to AI chatbots such as ChatGPT or Claude every day, and not every answer pleases them. Anthropic, the company that makes Claude, has now issued a new rule on how users treat the chatbot when they are unhappy with a reply.

The company says people who repeatedly behave cruelly or abusively towards the AI without any reason may face action. Users have since been asking whether this could mean their account being shut. It has been clarified that expressing displeasure at an AI's answer, or criticising it, will not by itself lead to an account being closed.

It is often seen that when an AI does not answer as a user wishes, many people start using foul language while talking to it. Experts say this can end up harming the users themselves. Consider a user who asks the AI a question and gets a wrong answer, asks the same question again and gets another wrong answer, and then, in anger, swears or uses abusive words. Could the account be closed? This is the question now being raised.

Anthropic has barred repeated abusive or improper behaviour towards its AI models in its policy. The company announced the new policy from October 8, 2026, and the rule will apply everywhere from November 12. The policy prohibits "sustained and needless abusive or cruel behavior", that is, abusive or cruel conduct without reason.

The company said the rule is meant only for serious cases in which users repeatedly treat the AI abusively and there is no clear purpose behind it.

This means that if Claude gives a wrong answer to a question and the user tells it the answer is wrong, that will not be treated as a violation. Users may disagree with Claude's answer, question it and ask it to give a better one. Testing the model, or having it write on serious and deep subjects for a story, has also been kept outside the scope of the rule.