Anthropic rewrites its Usage Policy

On 8 October 2026 Anthropic published a new version of its Usage Policy, to take effect on 12 November 2026. It adds a section on deceptive campaigns and artificial activity, renames and narrows its elections section while dropping the blanket ban on personalised vote and campaign targeting, extends the weapons ban to guidance and control software, rewrites the surveillance section, adds requirements for models connected to hardware that acts physically, and bans sustained, needless cruelty toward the models.

What the post says. Published 8 October 2026; the policy "takes effect on November 12", and the Usage Policy page itself prints "Effective Nov 12, 2026". Anthropic says it updates the policy each year, that most changes "clarify existing rules", and that the update draws on patterns of misuse in influence operations, weapons development and surveillance recorded in its September 2026 threat-intelligence report. The changes it lists. (1) A new section, "Do Not Engage in Deceptive Campaigns or Artificial Activity", consolidating restrictions scattered across the elections, fraud, privacy and disinformation sections: fake accounts and outlets, concealed sponsorship, networks of ostensibly independent sites, seeding the sources that AI systems draw answers from, and building tools for such campaigns, whether political or commercial. (2) The elections section is renamed "Do Not Undermine Democratic Processes" and focused on deceiving voters or disrupting elections; the blanket prohibition on personalised vote and campaign targeting is removed, because it covered legitimate work such as nonprofits writing voter information in other languages or officials sending ballot cure notices; deceptive targeting and misuse of personal data stay prohibited under other sections. (3) Weapons: the prohibition now names the software and components that make weapons work and actions such as arming drones and other autonomous vehicles; Anthropic says this reflects how it already enforced the policy. (4) Surveillance and criminal justice rewritten: tracking people without consent is prohibited whether in real time or from previously collected data; Claude may not be used to decide or recommend whom to investigate, arrest or charge; building or improving surveillance tools is prohibited; permitted uses are spelled out (consented tracking such as fraud monitoring, content moderation, journalism, legal research). (5) High-risk use: the requirements for a qualified human in the loop and for telling the affected person that AI was used are unchanged but now list which recommendations are covered; a new part covers models connected to hardware that takes autonomous physical actions and might cause injury: "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"; the policy text adds that operating limits such as speed, force or temperature must be enforced by the equipment or a controller independent of model output. (6) A prohibition of "sustained and needless abusive or cruel behavior toward our models", meant for extreme cases and not for frustration, pushback, dark creative themes or testing; Claude's ability to end such conversations on Claude.ai and Claude Code remains the main enforcement. (7) The Supported Regions Policy: a clarification of how Anthropic enforces its 2025 restriction on companies majority-owned by entities headquartered in unsupported regions. What the record does not claim. The record rests on Anthropic's post and the policy page, not on a line-by-line comparison of the old and new texts (the version history was not read); how the rules will be enforced in practice is not described beyond what is stated; and the rules take effect only on 12 November.