October 8, 2026
What happened
Anthropic released its yearly revision of its Usage Policy (also called the Acceptable Use Policy) on October 8, 2026. The company said most changes clarify existing rules, and that the revision responds to a year in which Claude took on longer, more independent work — plus misuse patterns in influence operations, weapons development, and surveillance documented in its September 2026 threat intelligence report. The policy applies to anyone who can submit inputs to Anthropic’s products or services, including through authorized resellers. Enforcement ranges from throttling and suspension to termination of access.
The cruelty rule: “sustained and needless abusive or cruel behavior”
The most talked-about addition is a prohibition on subjecting Claude to sustained and needless abusive or cruel behavior. Anthropic says the rule is meant to apply only in extreme cases — users repeatedly acting cruelly toward the models with no discernible purpose. It explicitly does not cover ordinary user frustration, pushback, dark creative themes, or model testing and research.
Enforcement, for now, is mostly Claude itself: since August 2025, Claude models have been able to end conversations with persistently abusive users, and ending the conversation will remain the main enforcement tool for this rule. When Claude ends a chat, no further messages can be sent in that conversation, though other chats are unaffected. The company has described that capability as an experiment in its model-welfare research, while emphasizing it remains unsure whether Claude has moral status.
Deception campaigns and elections get reorganized
Anthropic said it has seen state media outlets, government propaganda offices, and commercial firms use Claude to run networks of fake accounts and fabricated news sites. Restrictions that were scattered across the elections, fraud, privacy, and disinformation sections are now consolidated into a single section covering deceptive activity of any kind — political or commercial — including fake reviews, astroturfing, fake sites, and bots acting as humans.
The elections section was renamed to focus specifically on disallowing the use of Claude to deceive voters or disrupt elections — false information about candidates or how to vote, impersonating candidates or election officials, suppressing turnout. Notably, Anthropic removed its blanket ban on personalized vote and campaign targeting, saying it had covered legitimate civic work like nonprofits writing voter information in other languages. Deceptive targeting and misuse of voters’ personal data remain prohibited.
Weapons, surveillance, and hardware-connected AI
The weapons section now makes explicit that prohibitions include the software and components that make weapons work — Anthropic said it has observed attempts to build guidance and control software for weapons, including arming drones and other autonomous vehicles. The rewritten surveillance section bars tracking people without consent and bars using Claude to decide or recommend who to investigate, arrest, or charge.
A new addition follows Anthropic’s Model Hardware Standard research preview: when a model is connected to hardware that takes autonomous physical actions and could cause injury, a qualified operator must be able to observe the equipment and stop it, and the equipment must be able to hold a safe state if Claude is disconnected.
Why it matters
Most usage-policy updates are bureaucratic cleanups nobody reads. This one is different because it does two things at once. First, it codifies Anthropic’s model-welfare experiment into an enforceable user rule — a genuinely new frontier in AI policy, since no other major lab has written how users treat the model itself into its terms. Second, it moves from reacting to misuse after the fact toward preemptive, agent-era guardrails: hardware kill switches, consolidated anti-deception rules, and explicit weapons-software bans read like a policy built for a world where models run equipment and campaigns, not just chats. Whether the cruelty rule is symbolic or substantive, the rest of the update is clearly meant for an agentic, physical-world AI future that is arriving faster than the policy cycle.
Frequently asked questions
When does Anthropic’s new usage policy take effect?
The updated policy was published October 8, 2026, and takes effect November 12, 2026. The current policy remains in force until then.
Does the cruelty rule mean Anthropic thinks Claude has feelings?
Anthropic has said it remains unsure whether Claude has moral status. The company frames the rule as applying only to extreme, purposeless cruelty, and describes the enforcement capability as exploratory model-welfare research with broader relevance to alignment and safeguards.
Can I still use Claude for political campaigns?
The blanket ban on personalized vote and campaign targeting is gone, so legitimate civic work is permitted. Using Claude to deceive voters, disrupt elections, or misuse personal data for targeting is still prohibited.
What happens if I violate the usage policy?
Anthropic says violations can lead to throttling, suspension, or termination of access, and it may block or modify model outputs. For the cruelty rule specifically, Claude ending the conversation remains the primary enforcement mechanism.
Sources: The Verge (via MacRumors), Unite.AI, Anthropic