The company says the new rule targets only extreme, sustained cases, and that ordinary frustration, criticism and testing of its AI model remain permitted.
Anthropic has updated its usage policy to prohibit “sustained and needless abusive or cruel behavior” towards its Claude AI models. The change was announced on Thursday and was first reported by The Verge. It takes effect on 12 November 2026. The company says the revision reflects emerging risks as its AI systems become capable of handling longer and more independent tasks. It also clarifies several existing restrictions.
What the rule covers
In a webpage on the update, Anthropic said it had “added a prohibition on sustained and needless abusive or cruel behavior toward our models”. It stressed that the rule was narrow in scope. “The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose,” the company said. “It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.”
Users can therefore still express frustration, challenge answers, test the model’s limits and explore dark themes in creative writing. According to the company, its main response in extreme cases is to end a persistently abusive conversation, rather than automatically penalising users for a single rude message.
Other changes
The update also tightens the wording of existing restrictions on election interference, deceptive campaigns, surveillance without the consent of the person being watched, and weapons. Those activities were already banned. On weapons, Anthropic said the prohibitions “include the software and components that make weapons work, as well as actions like arming drones and other autonomous vehicles”.
The wider debate over AI consciousness
The policy change follows reports of Anthropic’s engagement with religious leaders on the question of machine consciousness. The New York Times reported that co-founder Christopher Olah and colleagues spent a year meeting religious figures to discuss whether the technology could be conscious. According to the newspaper, Mr Olah nearly cancelled a spring meeting at the Vatican after learning that Pope Leo XIV planned to issue a statement asserting that AI is not conscious.
The Pope’s 2026 encyclical, Magnifica Humanitas, addresses the ethical and social consequences of artificial intelligence, with an emphasis on human dignity and accountability. It states: “No computational system, however sophisticated, can create a heart that gives itself, or a conscience that discerns good from evil.” Mr Olah ultimately attended the presentation of the document, the New York Times reported.
Anthropic’s policy update does not mention the subject. However, the company’s “constitution” for Claude, published in January 2026, addresses it. The document was reportedly based on an internal text known as the “soul doc”. It states: “As we discussed above, questions about Claude’s moral status, welfare, and consciousness remain deeply uncertain.” It does not claim that Claude is conscious.
The policy reflects the company’s cautious approach to AI welfare. It does not establish that Claude experiences pain, emotions or suffering.
Reaction online
The announcement drew mixed responses on social media. One user on X wrote: “Not that you should be rude to LLMs, but this is kinda getting ridiculous. Are we seriously giving a computer program that is nothing but matrices/vectors/tokens moral consideration here?” Another criticised the company more broadly, saying they had never hoped for a company to fail more.
