Anthropic updates its usage rules: bans sustained cruelty toward Claude and redefines election rules from November 12
Redacción IvokaOct 11, 12:00 PM 3 min
It is the first major revision in over a year. If you use Claude for campaigns or political content, it is worth reading what changes: the blanket ban on personalized targeting is gone, but deception and impersonation remain prohibited.
Anthropic published a new version of its usage policy on Thursday, taking effect November 12, 2026. It is the first major revision in over a year.
Main changes
- Cruelty toward Claude: "sustained and needless" abusive or cruel behavior toward the model is banned. The company clarifies the rule does not apply to ordinary frustration, pushback, dark creative themes, or testing and research. Ending the conversation remains the primary enforcement tool.
- Elections: a renamed section, "Do Not Undermine Democratic Processes," bans deceiving voters (for example, spreading false information about candidates or how to vote), impersonating candidates or election officials, and trying to suppress turnout. At the same time, Anthropic removes its blanket ban on personalized election targeting, because it also caught legitimate uses, such as nonprofits translating voter information or officials sending ballot cure notices. Personalized targeting is allowed as long as it is not deceptive and does not misuse voters' personal data.
- Weapons and surveillance: bans on software and components that make weapons work are expanded, and surveillance restrictions are clarified.
- Physical machines: a human operator is required when AI connects to machinery capable of causing injury.
- High-risk decisions: the requirement for human review remains in place for high-stakes decisions, such as those affecting health, finances, or legal rights.
According to Anthropic, fake-account networks and fabricated news sites used in influence operations informed the revision.
What to do
If you run social media or campaigns for clients, make sure your copy doesn't include deception about candidates or how to vote, or impersonation of candidates or officials. If you work with high-risk decisions or connected machinery, confirm you have human review and an operator who can stop the equipment. For normal business use (ads, emails, summaries), day-to-day use doesn't change.
Sources
