Anthropic Tightens Usage Rules to Curb Bot Abuse and High-Risk Applications

Anthropic has updated its usage policy to prohibit "sustained and needless abusive or cruel behavior" toward its Claude AI models, as The Register first reported. The new rules, which take effect November 12, target extreme cases of repeated cruelty with no discernible purpose, though Anthropic notes the policy will not apply to model testing, research, dark creative themes, or common user frustration.

To enforce these etiquette rules, Anthropic will rely on a feature introduced in August 2025 as part of "model welfare" research, which allows Claude to terminate conversations with users who are persistently abusive.

The anti-cruelty measure is part of a broader policy overhaul addressing more consequential security risks. The Register reports that Anthropic has tightened restrictions on deceptive influence campaigns after finding commercial organizations, government propaganda offices, and state media outlets used the chatbot to run fabricated news sites and fake accounts. While Anthropic has removed a blanket ban on personalized political targeting to allow for activities like nonprofit voter information translation, it continues to prohibit the misuse of personal information and deceptive targeting.

Additionally, Anthropic clarified that its ban on weapons development now explicitly includes software and components, specifically targeting the creation of autonomous vehicles and armed drone guidance systems. The company also prohibited using Claude to develop surveillance tools or to recommend individuals for police investigation, arrest, or charging. For users connecting Claude to hardware capable of causing injury, Anthropic now requires a qualified human operator to monitor the equipment and ensure it remains in a safe state if the AI connection is lost.