Anthropic has updated Claude’s usage policy after more than a year. The new rules expand restrictions in the areas of election interference, propaganda, surveillance, and weapons development. The policy also prohibits “persistently and unnecessarily aggressive or cruel behavior” toward Claude; however, users’ anger, objections, dark-themed creative works, and security research are exempt. In such cases, the primary sanction will be ending the conversation; no information was provided regarding account suspension.
Among the changes are bans on spreading content through fake accounts, creating fake news organizations, and misleading voters. Impersonating candidates or election officials is also banned under the heading “Do not undermine democratic processes.” The weapons-related ban also covers control software and the weaponization of unmanned aerial vehicles. Tracking individuals without consent and use in law enforcement decisions have been prohibited. A qualified operator requirement has been introduced for physically autonomous systems; the rules may be relaxed in some government contracts.
Background
Anthropic is not a new name in the FikirPilot archive: we have published 39 stories mentioning this name in the last 90 days; the most recent is dated October 8, 2026.