Anthropic updates AI usage policy, disallowing abuse and easing campaign targeting rules
AI company Anthropic has updated its usage policy, which takes effect November 12. The changes prohibit "sustained and needless abusive or cruel behavior" towards its AI models but loosen restrictions on political campaign targeting.
Artificial intelligence company Anthropic has updated its usage policy, with new terms prohibiting "sustained and needless abusive or cruel behavior" towards its AI models, effective November 12. The company stated the rule applies only in extreme cases where users repeatedly act cruelly toward models with no discernible purpose, excluding user frustration, testing, or research.
The updated policy also removes a blanket prohibition on personalized vote and campaign targeting. Anthropic explained this change aims to allow legitimate civic work, such as non-profits providing voter information in multiple languages. However, voter targeting relying on deception or misuse of personal data remains banned under stricter rules for deceptive campaigns.
Anthropic has also clarified prohibitions regarding weapons and surveillance. The policy now explicitly bans using Claude for weapon guidance and control software, and for arming drones. Tracking individuals without consent, or making decisions about investigations, arrests, or charges, is also prohibited. Consent-based tracking for purposes like fraud monitoring or journalism remains permitted.
These policy shifts reflect Anthropic's existing enforcement practices. The company previously faced scrutiny from the U.S. Department of Defense for refusing to drop bans on mass surveillance and fully autonomous lethal decision-making, leading to a dispute that involved national security designations.