Anthropic bans abusive behavior toward Claude, tightens election and weapons rules
In short
Anthropic has updated its usage policy to ban "sustained and needless abusive or cruel behavior" toward its AI models, saying the rule applies only to extreme cases where users repeatedly act cruelly with no discernible purpose. The update, which takes effect on November 12, also adds rules on deceptive campaigns, election interference, weapons software and surveillance.
Read the full story 3 min read
Anthropic has updated its usage policy to prohibit "sustained and needless abusive or cruel behavior" toward its AI models, as part of broader changes announced on October 8. The new policy takes effect on November 12, MacRumors and Business Insider reported. The Verge said it is the first change to the usage policy in over a year, while Engadget described it as Anthropic's annual usage policy update. [ 1 , 2 , 3 , 4 , 5 , 6 ]
Anthropic said the rule "is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose," and that it does not apply to "common versions of user frustration, pushback, dark creative themes, or model testing and research." Ending conversations remains the "primary enforcement mechanism," The Verge reported. MacRumors said that when Claude ends a chat, no further messages can be sent in that conversation, but other chats are not affected. Anthropic did not comment on whether there would be further enforcement, such as user bans, according to The Verge, and CBS News said the company did not immediately respond to questions on what constitutes abusive or cruel behavior. [ 1 , 4 , 6 ]
Anthropic first let Claude end conversations with "persistently harmful or abusive" users in August 2025 as part of research into model welfare. MacRumors reported that Anthropic said at the time it was unsure whether Claude has moral status, and that it found Claude Opus 4 had a "robust and consistent aversion to harm." [ 1 , 4 ]
The update also gathers existing restrictions into a ban on deceptive commercial or political campaigns, including "efforts to obscure who is behind a message or amplify content through fake accounts or posts." Anthropic said it found state media outlets, government propaganda offices and commercial firms using Claude to run networks of fake accounts and fabricated news sites, MacRumors reported. An election section titled "Do Not Undermine Democratic Processes" bars voter deception and election disruption, which Engadget linked to the midterms. After multiple attempts to use Claude to develop guidance and control software for weapons, the ban on weapons development now covers "software and components that make weapons work" and actions like arming drones, The Verge reported. The update also addresses surveillance. [ 1 , 2 , 3 , 4 , 7 ]
Engadget said the change follows a viral "AI torture chamber" project, which Anthropic did not explicitly mention, and reports that Anthropic has met religious and philosophical leaders, including at the Vatican. TechCrunch reported Anthropic had sought collaborations with religious scholars, in some cases seeking to convince them Claude could be conscious or have a soul. Engadget quoted independent journalist Kat Tenbarge as saying Big Tech companies are "going to moderate violence against AI before they ever moderate violence against women and minorities." Business Insider reported critics said the rule risks treating an AI chatbot too much like a person, while supporters said it could make sense even if Claude is not conscious. CBS News cited an essay by Microsoft AI CEO Mustafa Suleyman, published last month, arguing that Anthropic is effectively "training Claude that it may be conscious," which he said would make AI harder to contain. [ 2 , 3 , 5 , 6 ]
Why it matters
The change makes repeated, purposeless cruelty toward an AI model a violation of Anthropic's usage policy. Business Insider reported critics said this risks treating an AI chatbot too much like a person, and CBS News placed it in a wider debate about AI consciousness. The deceptive-campaign rules respond to misuse Anthropic says it found, including networks of fake accounts and fabricated news sites.
Key facts
- Anthropic's updated usage policy prohibits "sustained and needless abusive or cruel behavior" toward its AI models. [ 1 , 2 , 3 , 4 , 5 , 6 ]
- Anthropic said the rule does not apply to common user frustration, pushback, dark creative themes, or model testing and research. [ 1 , 2 , 4 , 6 ]
- Ending conversations remains the primary enforcement mechanism; Claude has been able to end conversations with persistently abusive users since August 2025. [ 1 , 4 ]
- The new policy takes effect on November 12. [ 4 , 5 ]
- The update adds a section on deceptive campaigns, such as running fake accounts or fabricated news outlets. [ 1 , 2 , 4 ]
- An election section titled "Do Not Undermine Democratic Processes" bars deceiving voters and disrupting elections. [ 2 ]
- The weapons ban is extended to "software and components that make weapons work" and to actions like arming drones. [ 1 ]
Confirmed by several sources
- Anthropic banned "sustained and needless abusive or cruel behavior" toward its AI models in a usage policy update. [ 1 , 2 , 3 , 4 , 5 , 6 ]
- Anthropic said the ban applies only in extreme cases where users repeatedly act cruelly toward its models with no discernible purpose. [ 1 , 2 , 4 , 6 ]
- Claude could already end conversations with persistently harmful or abusive users. [ 1 , 2 , 3 , 4 , 6 ]
- The policy takes effect on November 12. [ 4 , 5 ]
- The update includes new rules on deceptive campaigns and on deceiving voters or disrupting elections. [ 1 , 2 , 3 , 4 , 7 ]
- The update also covers weapons development and surveillance. [ 1 , 2 , 7 ]
Still unclear
- Whether enforcement will go beyond ending conversations, for example user bans. The Verge reported Anthropic did not comment on further enforcement mechanisms.
- What exactly counts as "abusive" or "cruel" behavior toward Claude. CBS News reported Anthropic did not immediately respond to a request for comment on this.
- Whether the change is linked to the viral "AI torture chamber" project. Only Engadget made the connection and said Anthropic did not explicitly mention the project.
- Which Claude models can end abusive conversations. CBS News referred to Claude Opus, while MacRumors referred to Claude models in general.
What local media are saying
Timeline, local time
- October 8: The Verge reports Anthropic's usage policy update banning abusive or cruel behavior toward Claude. [ 1 ]
- October 8: TechCrunch reports the update, including election, weapons and surveillance rules. [ 2 ]
- October 8: MacRumors reports the policy takes effect on November 12. [ 4 ]
- October 9: CBS News reports Anthropic did not immediately respond to questions about what counts as abusive behavior. [ 6 ]