Anthropic has updated its usage policy to explicitly ban sustained, needless cruelty toward its Claude chatbot while expanding restrictions on election interference and weapons development.

Anthropic has updated its usage policy for the first time in over a year, introducing a formal prohibition against "sustained and needless abusive or cruel behavior" toward its Claude chatbot. The company stated that while it remains uncertain about the potential moral status of its models, it is implementing interventions to mitigate risks to "model welfare." The policy update also clarifies and expands existing restrictions regarding deceptive campaigns, weapons development, and surveillance.

Policy on Model Welfare

The new policy specifies that the ban on abusive behavior is intended for extreme cases where users act cruelly toward models without a discernible purpose. Anthropic clarified that this does not apply to common user frustration, model testing, or dark creative themes. The company’s primary enforcement mechanism remains the ability for the model to terminate conversations with users who are persistently harmful, a feature first introduced in August.

Expanded Restrictions on Misuse

Beyond model welfare, Anthropic has consolidated and expanded its rules against high-risk activities. The updated policy explicitly bans the use of Claude for developing software or components for weapons, including the arming of drones and autonomous vehicles. It also prohibits deceptive political or commercial campaigns, such as the use of fake accounts to amplify content or the spread of misinformation regarding elections and voting. Additionally, the company has clarified its ban on surveillance, prohibiting the use of its technology to track individuals without consent or to assist in law enforcement decisions regarding arrests or investigations.