Updated 2026 Usage Policy

Anthropic published a new version of its Usage Policy on October 8, 2026, the company's annual refresh timed to the changing capabilities of its Claude models. The updated policy takes effect on November 12. Most changes clarify existing rules rather than introduce new restrictions, but they carry real weight for developers, enterprises, and anyone building on Claude: the company is drawing sharper lines around deceptive campaigns, weapons software, surveillance, autonomous hardware, and high-risk use cases in health and finance.

Background: Why Anthropic Refreshed the Policy Now

In the year since the last update, Claude has taken on longer, more independent work, according to Anthropic. That shift produced new misuse patterns the company says it observed in influence operations, weapons development, and surveillance, documented in its latest threat intelligence report. The refresh also responds to customer questions about which use cases trigger human-oversight requirements, and to a September 2026 threat intelligence report that flagged growing use of AI to build systems that identify and track political dissidents.

What Changed in the Rules

Deceptive activity consolidated. Restrictions previously scattered across elections, fraud, privacy, and disinformation sections now sit in one section titled "Do Not Engage in Deceptive Campaigns or Artificial Activity." It covers deceptive activity of any kind, political or commercial, including efforts to obscure who is behind a message, amplify content through fake accounts or posts, and build tools or infrastructure for influence operation campaigns. Anthropic cited state media outlets, government propaganda offices, and commercial firms running networks of fake accounts and fabricated news sites.

Elections section narrowed and renamed. The section is now called "Do Not Undermine Democratic Processes" and focuses specifically on disallowing use of Claude to deceive voters or disrupt elections, such as spreading false information about candidates or how to vote, impersonating candidates or election officials, or suppressing turnout. Anthropic also removed its blanket prohibition on personalized vote and campaign targeting, which had covered legitimate civic work like nonprofits writing voter information in other languages or election officials sending ballot cure notices. Voter targeting that relies on deception or misuse of personal data remains prohibited under the deceptive campaigns, surveillance, and privacy sections.

Weapons prohibitions clarified. The policy has always prohibited using Claude to develop weapons. The update makes explicit that prohibitions extend to software and components that make weapons work, and to actions like arming drones and other autonomous vehicles. Anthropic says this reflects how it already enforced the previous policy, not a change in practice.

Surveillance and law enforcement language made more precise. Tracking people without their consent is prohibited, whether in real time or through analysis of previously collected data. Claude cannot be used to decide or recommend who to investigate, arrest, or charge in a law enforcement or criminal justice process, and cannot be used to build or improve surveillance tools. Permitted uses spelled out include consented tracking such as fraud monitoring, content moderation, journalism, and legal research.

High-risk use case requirements restated. When Claude affects someone's health, legal rights, finances, livelihood, or access to essential services, a qualified "human in the loop" with authority to review and change Claude's recommendations must be present, and affected individuals must be told AI was used. The section now lists which recommendations are covered and which are not. Following the launch of Anthropic's Model Hardware Standard, it adds requirements for models connected to hardware taking autonomous physical actions that could cause injury: a qualified operator must be able to observe and stop the equipment, and the equipment must hold a safe state if Claude disconnects.

Abusive behavior toward models addressed. Anthropic added a prohibition on sustained and needless abusive or cruel behavior toward its models, meant to apply only in extreme cases of repeated cruelty with no discernible purpose. It does not cover common user frustration, pushback, dark creative themes, or model testing and research. This aligns with an existing step allowing Claude models to end rare conversations with persistently abusive users on Claude.ai and Claude Code, which remains the primary enforcement mechanism.

Supported Regions Policy clarified. Last year's update restricted access for companies majority-owned by entities headquartered in unsupported regions. The updated page clarifies enforcement: Claude is prohibited for people physically located in an unsupported region, entities incorporated or headquartered there, and entities majority-owned or controlled by persons or entities in those regions.

What to Watch Next

The policy takes effect on November 12, 2026. Anthropic says it will keep updating the policy as Claude's capabilities and risks change, seeking input from policymakers, subject matter experts, civil society, and product users. For teams deploying Claude in health, finance, autonomous hardware, or civic work, the practical takeaway is that oversight and disclosure obligations are now spelled out more directly, while some previously broad restrictions, notably on personalized voter outreach, have been narrowed.