Anthropic has released a revised Usage Policy for Claude, its yearly rewrite of the rules that govern how customers and users may deploy the company's models. The company says most of the changes clarify rules it already enforced, but several sections are reorganized around misuse it has observed over the past year, and a few obligations are new. The updated policy takes effect on November 12, 2026.
The most visible change is a standalone section titled "Do Not Engage in Deceptive Campaigns or Artificial Activity." Anthropic says restrictions on fake account networks and fabricated news sites were previously scattered across its election, fraud, privacy and disinformation clauses. The new section gathers them in one place and applies to political and commercial deception alike. It covers hiding who is behind a message, inflating reach with fake accounts or posts, and building the tools or infrastructure used to run influence campaigns.
Election rules move in the other direction. The section is renamed "Do Not Undermine Democratic Processes" and now focuses on deceiving voters or disrupting elections, for example by spreading false voting information, impersonating candidates or officials, or suppressing turnout. Anthropic has dropped its blanket ban on personalized vote and campaign targeting, saying it blocked legitimate civic work such as nonprofits translating voter information or election offices sending ballot cure notices. Deceptive targeting and misuse of voter data stay prohibited under other sections.
Weapons and surveillance language is sharper, though Anthropic says enforcement is unchanged. The weapons ban now explicitly includes guidance and control software, weapon components, and arming drones or other autonomous vehicles. The surveillance section bars tracking people without consent, whether live or from stored data, prohibits using Claude to decide or recommend who should be investigated, arrested or charged, and bans building or improving surveillance tools. It also lists uses that remain allowed, including consented tracking such as fraud monitoring, content moderation, journalism and legal research.
The high-risk use section keeps its existing requirements: a qualified human must be able to review and override Claude's recommendations in areas such as health, legal rights and finance, and affected people must be told AI was involved. What is new is a set of controls for hardware that can take autonomous physical actions and might cause injury. A qualified operator must be able to watch the equipment and stop it, and the machine must hold a safe state if Claude disconnects. Anthropic ties this to its recently launched Model Hardware Standard.
The policy also adds a narrow prohibition on "sustained and needless" abusive or cruel behavior toward Claude. Anthropic says it targets only extreme, purposeless cases, not frustration, dark creative themes or model testing, and that its main enforcement tool remains Claude's existing ability to end persistently abusive conversations on Claude.ai and Claude Code. Separately, the Supported Regions Policy now spells out that access is barred for people located in unsupported regions, for entities incorporated or headquartered there, and for entities majority-owned or controlled from those regions.
For businesses building on Claude, the practical work is a compliance review before November 12, especially for robotics or lab automation, civic outreach tools, and any product that scores or monitors people. The policy text itself is the authoritative reference; the announcement summarizes intent, and how the new clauses are applied in edge cases will only become clear through enforcement.