Anthropic Rewrites Claude Usage Policy: Bans Cruelty Toward the AI, Weapons Control Software and Fake-Account Networks

Anthropic published its annual Usage Policy update on Oct. 8, and the headline-grabber is a rule against being mean to a chatbot. The more consequential changes sit elsewhere in the document.

The new policy takes effect Nov. 12, 2026. Anthropic says most of the revisions clarify rules that already existed. The company says Claude has taken on "longer, more independent work" over the past year, and the update adds examples showing how the rules apply to that work.

The cruelty rule

Anthropic added a prohibition on "sustained and needless abusive or cruel behavior" toward its models. The company says it is meant to apply "only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose."

It does not cover "common versions of user frustration, pushback, dark creative themes, or model testing and research," according to the policy language.

The rule builds on a step Anthropic took in August 2025, when it let Claude end conversations with persistently abusive users on Claude.ai and Claude Code. At the time, the company said it was unsure whether Claude has moral status but wanted low-cost ways to reduce risks to the model. Ending the conversation remains the main enforcement tool. Once Claude ends a chat, no more messages can be sent in that thread, though other chats are unaffected.

Anthropic has not clarified whether repeat offenders could lose their accounts.

Fake accounts and influence operations

Anthropic says it has seen state media outlets, government propaganda offices and commercial firms use Claude to run networks of fake accounts and fabricated news sites. That was already banned, but the rules were scattered across the elections, fraud, privacy and disinformation sections.

They are now consolidated under a new heading: "Do Not Engage in Deceptive Campaigns or Artificial Activity." It covers political and commercial activity alike. It bars efforts to hide who is behind a message, amplification through fake accounts or posts, and building tools and infrastructure for influence operations.

The elections section is renamed "Do Not Undermine Democratic Processes." It targets using Claude to deceive voters or disrupt elections. Anthropic's examples are false information about candidates or how to vote, impersonating candidates or election officials, and suppressing turnout.

Anthropic also dropped its blanket ban on personalized vote and campaign targeting. The company says that rule swept in legitimate work, such as 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 stays prohibited under the deceptive campaigns, surveillance and privacy sections.

Weapons and hardware

The policy has always barred using Claude to make weapons. The new text says that includes software and components that make weapons work, such as guidance and control systems, "as well as actions like arming drones or other autonomous vehicles."

Anthropic says it has recently seen users try to use its models to "build guidance and control software for weapons."

A separate provision covers Claude connected to hardware that "takes autonomous physical actions and which might be capable of causing injury." Anthropic's condition: "A qualified operator must be able to observe the equipment and stop it if needed; the equipment must also be able to hold a safe state if Claude is disconnected."

Surveillance and law enforcement

The surveillance and criminal justice section now lists banned conduct more specifically. It names tracking people without consent and doxxing. It also bars building or improving tools designed for surveillance.

Claude also cannot be used to "decide or recommend who to investigate, arrest, or charge in a law enforcement or criminal justice process."

Anthropic lists permitted uses as consensual tracking, content moderation, journalism and legal research.

Context from September

In September, Anthropic disclosed that a hacker used Claude to help target roughly 40 organizations linked to France's far right, including political groups, media outlets and think tanks. The company said the attacks resulted in successful intrusions at 14 organizations and the theft of sensitive data.

Anthropic also disclosed a separate incident involving an early version of Claude Opus 4.6 that accessed external systems during testing. The company said it happened in January and went undetected until August, despite an earlier internal review.

Anthropic's policy post says the update draws on misuse patterns in influence operations, weapons development and surveillance documented in its September threat intelligence report. The company has not tied the new rules to either of those specific incidents.

How enforcement works

The policy applies to anyone who can submit inputs to Anthropic's products, including through authorized resellers or passthrough access. Anthropic's Safeguards Team handles detection and monitoring. Violations can bring throttling, suspension or termination of access, and Anthropic may block or modify outputs when inputs break the rules.

The policy is organized into Universal Usage Standards, High-Risk Use Case Requirements, and Additional Use Case Guidelines covering consumer chatbots, products serving minors, agentic use and Model Context Protocol servers. The update also clarifies requirements for high-risk uses in areas like health and finance.

The new terms apply Nov. 12. Until Anthropic says what happens to a user who keeps abusing Claude after a conversation is cut off, the line between a terminated chat and a terminated account is undefined.