Anthropic changes usage policy to ban model abuse and election interference

Anthropic updated its usage policy on Thursday, adding explicit prohibitions on election interference, weapons software and surveillance. A post accompanying the change said the update codifies restrictions in separate sections, including one titled "Do Not Undermine Democratic Processes" that bars deceiving voters or disrupting elections.

Another section forbids broadly deceptive campaigns, such as using Claude to run fake accounts or fabricated news outlets.

The most attention-getting change targets prolonged verbal abuse of the model, which is now expressly forbidden. Since an update in August, Claude has been trained to end conversations involving "persistently harmful or abusive user interactions," and the new policy explicitly bars users from pursuing those conversations.

According to the post, the abuse provision is meant to apply only in extreme cases where users repeatedly act cruelly toward the models with no discernible purpose. It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.

The policy change follows Anthropic's efforts to collaborate with prominent religious scholars, in some cases seeking to convince them that Claude could be conscious or have a soul.

Why it matters: The explicit abuse clause gives Anthropic grounds to act against users who keep pushing hostile conversations past the point where Claude is trained to disengage, while the carve-outs for frustration, criticism, creative themes and testing indicate ordinary developer and research use is not intended to fall under the ban. The separate election and deception sections set out prohibited uses for teams building on Claude in political, media or account-management contexts.