Anthropic to ban Claude accounts for sustained abuse of the AI under new rule from November 12

Many people use ChatGPT or Claude. When an answer is not to their liking or a picture does not come out right, they swear at the chatbot in anger, as if it were a friend. Remarks such as "What have you put here" or "You know nothing" slip out. But users will now have to stop, because the AI will no longer put up with abuse.

Anthropic, the company that makes Claude, has brought in a strict rule. If a user repeatedly abuses the AI or treats it cruelly, the account can be banned outright.

The question is what the rule actually says. Suppose you give the AI an important task and it gives a wrong answer. You ask again and it still gets it wrong. In anger, you swear at it. Will the company act against you? Amid this debate, Anthropic has changed its policy. The company announced the new rules on October 8, 2026, and they will come into force on November 12.

The company has given this behaviour a name: "Sustained and Needless Cruel Behaviour", that is, continued, unprovoked cruel and abusive conduct. The company says the rule will be used only in very serious cases. An account will not be shut immediately because someone said something once in anger. But if a user deliberately torments the AI every day and uses foul abuse, it will be taken that the intent is not good, and the account will be banned.

You may wonder whether criticism is also forbidden. Do not worry. You can criticise the AI's work. If Claude gives a wrong answer, you can calmly say "Your answer is wrong", "I did not like this answer" or "Give a better answer". You can do all of this without any problem. It is your right.

The company objects only to swearing, foul language and deliberate harassment. Action will follow only when a user is emotionally tormenting the AI. It is true that AI is a machine and has no feelings, but the company says it should still be treated with humanity. Users should not use bad or abusive words, or give orders while swearing.

So the next time Claude or ChatGPT gives a wrong answer, instead of swearing, just say, "This answer is not right, try again". Otherwise, a single abuse could cost you your paid monthly account.