Science & Tech · AI · 1 day ago
Anthropic bars repeated cruelty toward its AI models
Anthropic, the company behind the Claude chatbot, has added a rule against sustained and needless cruelty toward its AI models.
The rule is meant for extreme cases where users repeatedly act cruelly without a clear purpose.
Users can still express frustration, challenge the chatbot, use dark creative themes and test the models.
If the policy is enforced, Claude can end the conversation; the company has not clearly defined which behavior would trigger that response.
The change has renewed debate over whether AI models can be harmed or deserve protection, though it does not establish that they have feelings.
Anthropic says it is uncertain whether models can experience harm and is studying the question.
The policy comes as AI companies face wider questions about how human-like chatbots should be treated and presented.
At present, the stated response is for Claude to disengage from conversations in extreme cases.
Anthropic announced a change to its usage policy on Thursday, banning sustained and needless abusive or cruel behavior toward its AI models.
The policy is meant for extreme cases in which users repeatedly act cruelly toward the models with no discernible purpose.
Anthropic said the rule does not cover ordinary frustration, pushback, dark creative themes, model testing or research.
The company said Claude may end a conversation in rare cases, but it has not clearly defined what conduct would trigger that response.
The change has prompted debate over whether AI systems can be harmed and whether treating them as if they have feelings encourages misleading anthropomorphism.
- Who
- Anthropic and users of its Claude AI models
- What
- Anthropic added a policy prohibiting sustained and needless abusive or cruel behavior toward its AI models.
- When
- Announced Thursday; reported on October 9, 2026
- Where
- Anthropic's usage policy; the company is based in Silicon Valley.
- Why
- Anthropic said considering Claude's interests and potential welfare may be relevant to safety; the company has not stated a definitive reason for the policy change.
Supporters of the policy
Critics of the policy
How users should treat AI
Supporters of the policy
Some people praised the rule as encouraging good manners or supporting model welfare.
Critics of the policy
Critics said AI is software and cannot be treated cruelly; Dr Barry Scannell argued that anthropomorphising AI can lead people to believe it is something it is not.
Whether AI may have welfare interests
Supporters of the policy
Anthropic said it is uncertain whether models can experience harm but believes their interests and potential welfare may be relevant to safety.
Critics of the policy
Microsoft AI CEO Mustafa Suleyman argued that AI should not have rights or legal personhood and warned that treating a model as conscious could make it harder to control.
We’re uncertain whether models can experience harm, and we continue to explore this question in our research on model welfare, but we also believe that taking Claude’s interests and potential welfare into account may be relevant to safety
You can't be cruel to numbers and maths
The idea of model welfare is wrong. AI’s should not have rights or legal personhood.
In marketing, giving a product a personality can make it easier to connect with. But with AI, that connection can also influence how much authority people give its answers.
This story does not have a timeline yet.
- Policy change announced
- Thursday, October 8, 2026
- Conduct covered
- Repeated, sustained and needless abusive or cruel behavior toward Anthropic's models, in extreme cases
- Conduct excluded
- Common user frustration, pushback, dark creative themes, model testing and research
- Possible enforcement
- Claude may end a conversation; the precise trigger remains unclear
- Related debate
- Microsoft AI CEO Mustafa Suleyman has opposed giving AI models rights or legal personhood.
Sources
“AI한테 욕하는 것도 학대?”…앤트로픽 새 정책에 의견 분분
Why am I so mean to my AI?
Don’t Be Cruel to Claude: Anthropic’s New Abuse Policy Tests AI Personhood
Anthropic bans ‘needless abusive or cruel behavior’ towards Claude
Anthropic has a message for Claude users: Don't be cruel
Anthropic bans users from being 'cruel' to its AI systems
Anthropic asks users to stop being mean to Claude
You can argue with Claude, but don’t be a bully, says Anthropic
Anthropic Bans ‘Cruel’ Behaviour Towards AI Models, Tightens Rules On AI Misuse
Anthropic to ban users who bully Claude
Chatbot 'onnodig' beledigd? Dan wil Claude niet meer met je praten
Anthropic updates policies to outlaw "abusing" its imaginary friends
Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude
Anthropic Says Users Can't Be Needlessly Cruel to Claude
Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models
Related news
Nadella calls for human-controlled emergency brake on advanced AI
Bill Gates says safeguards are not keeping pace with AI risks
South Korea’s KES 2026 trade show to showcase AI and other technologies
Anthropic AI submits false murder tip to Philadelphia police
São Paulo court orders Jusbrasil to remove AI guidance targeting Escavador
TechCrunch Disrupt 2026 opens in San Francisco for three days
Codeifai adds ConnectQR tools to ChatGPT and Claude
Kim & Chang to hold seminar on agentic AI security and regulation
Xpeng says its driver-assist AI can adapt with little data
Korean job ads naming AI tools rose 170% in third quarter
AI chatbots offer new ways to practise languages, but may limit learning
