Loading page…
The November 12 policy draws a narrow boundary around abusive conversations, preserves bans on deceptive voter targeting and adds stop controls for potentially dangerous hardware.
Listen to this story
Anthropic’s revised Usage Policy, published October 8 and effective November 12, gives clearer boundaries for using Claude in civic, political, and operational settings. It permits personalized election outreach when it is not deceptive or based on misuse of voters’ personal data, while explicitly prohibiting sustained, purposeless cruelty toward the model. The update also consolidates rules against influence campaigns and clarifies safeguards for weapons, surveillance, and injury-capable hardware—important distinctions for teams deploying Claude in higher-stakes workflows.
The cruelty rule is limited to persistent abuse without a discernible purpose; frustration, dark creative work, testing, and research are excluded.
Claude may support legitimate targeted voter information, but not voter deception, impersonation, false voting claims, or turnout suppression.
Anthropic says fake-account networks and fabricated news sites used in influence operations informed the policy revision.
Claude users will face an explicit ban on sustained, purposeless cruelty toward the AI, while personalized election campaigning will no longer be prohibited across the board. Anthropic published the revised Usage Policy on October 8, 2026, with a November 12 effective date. The revision also addresses deceptive campaigns, weapons software, surveillance and potentially dangerous autonomous hardware.
Not every provision represents a new restriction. Anthropic says most changes explain existing rules more clearly as Claude takes on longer, more independent work. The company says the revision draws on misuse it observed over the past year in influence operations, weapons development and surveillance. Some provisions change policy; others spell out how existing restrictions apply.
The new abuse provision targets extreme cases in which users repeatedly act cruelly toward Anthropic’s models without a discernible purpose. Anthropic says ordinary frustration, pushback, dark creative themes, model testing and research are outside the rule’s scope. The restriction is not a general requirement to agree with Claude or avoid difficult subject matter.
The enforcement mechanism already exists: Claude can end rare conversations with persistently abusive users on Claude.ai and Claude Code. Anthropic says ending those exchanges will remain its primary enforcement mechanism. The revision turns conduct already addressed through that product behavior into an explicit user-facing prohibition.
It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.
Anthropic, October 8 Usage Policy announcement
Anthropic removed its blanket prohibition on personalized vote and campaign targeting because it also covered legitimate civic work. Its examples include nonprofits writing voter information in other languages and election officials sending ballot-cure notices. Targeting that relies on deception or misuse of voters’ personal data remains prohibited under other policy sections.
The renamed election section, Do Not Undermine Democratic Processes, focuses on deceiving voters or disrupting elections. It prohibits false information about candidates or voting, impersonation of candidates or election officials, and attempts to suppress turnout.
A separate consolidated section covers deceptive activity whether political or commercial. Previously scattered restrictions now sit together: hiding who is behind a message, amplifying material through fake accounts or posts, and building infrastructure for influence campaigns. Anthropic says it has seen state media, government propaganda offices and commercial firms use Claude for fake-account networks and fabricated news sites.
The operational provisions distinguish prohibited work from safeguards required for permitted uses. Anthropic says the weapons and surveillance wording makes existing enforcement more explicit, rather than changing it. The hardware requirements address models connected to equipment that acts autonomously and might cause injury.
For high-risk uses affecting health, legal rights, finances, livelihoods or access to essential services, the human-review requirement remains unchanged. A qualified person must have authority to review and change Claude’s recommendations, and affected people must be told AI was used. The rewritten section lists which recommendations are covered and which are not.
Story updates
Anthropic’s decision to prohibit sustained, needless cruelty toward Claude is drawing fresh scrutiny before it takes effect on November 12, 2026. In BBC interviews published October 15, experts split over whether the rule is a modest precaution or encourages misplaced concern for software. The practical consequence is narrower: Claude can end persistently abusive conversations.
Anthropic’s position began with an unresolved question, not a declaration that Claude has feelings. In a February interview with The New York Times, CEO Dario Amodei said the company did not know whether its models were conscious, or even exactly what consciousness would mean for a model. He nevertheless said Anthropic was open to the possibility.
That uncertainty underpins the company’s work on model welfare: whether an AI system could have wellbeing worth protecting. Anthropic describes questions about AI consciousness and wellbeing as deeply uncertain. The BBC’s feature reports no confirmed scientific finding that current AI systems are conscious. The policy therefore concerns how people should behave under uncertainty, rather than implementing a demonstrated discovery about machine suffering.
Anthropic published its Usage Policy update on October 8. The new provision prohibits “sustained and needless abusive or cruel behavior” toward its models. The company says it is intended only for extreme cases: repeated cruelty with no discernible purpose. That wording makes persistence and purpose central to the restriction, rather than simply whether a user speaks harshly to a chatbot.
The announcement explicitly leaves several kinds of interaction outside the prohibition:
Ordinary user frustration and pushback.
Dark themes in creative work.
Model testing and research.
Enforcement builds on a capability Anthropic says it already introduced. Claude models can end rare conversations with persistently abusive users on Claude.ai and Claude Code. The company says ending those interactions will remain the primary enforcement mechanism. The November change thus adds an explicit usage restriction around behavior the products already have a way to interrupt.
The assumption is that everybody thinks that the models are, in fact, conscious, or are very likely to be conscious. And instead, [the call for AI welfare] is coming from uncertainty and caution and humility,
Sebo, a leading advocate of AI welfare, told the BBC that he considers the probability of present-day AI consciousness low, but not zero. He singled out Anthropic’s cruelty policy as a small intervention with few potential downsides. His argument is precautionary: a limited response need not require certainty, just as people take modest precautions against unlikely but serious weather events.
Critics see a different risk in the language itself. In an October 14 opinion column, Daniel McCarthy argued that calling mistreatment of software “cruelty” encourages people to treat AI as conscious. He accepted that Anthropic can prohibit abuse of its products, but rejected the moral framing. His objection concerns what the rule teaches users to believe, not the company’s authority to set conditions.
The disagreement also reaches corporate accountability. University of Washington computational linguist Emily M Bender told the BBC that concerns about AI consciousness and wellbeing can shift accountability away from the people and companies running software. Sebo acknowledged related dangers: granting AI rights could help companies consolidate power, and AI welfare could conflict with human and animal interests.
For the November rollout, Anthropic has committed to a much smaller measure than that wider rights debate: a conduct rule backed primarily by ending conversations. The remaining practical judgment is how to distinguish purposeless, persistent abuse from the frustration, difficult fiction and testing the company expressly permits. Its announcement describes that boundary qualitatively, as extreme repeated cruelty without a discernible purpose.
anthropic.comLoading discussion...
Join the conversation
Explain whether this is a reasonable boundary or an unnecessary restriction.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.