Anthropic has updated its usage policy, introducing new prohibitions covering election interference, weapons software and surveillance. The most notable change, announced on Thursday, is an explicit ban on prolonged verbal abuse of its Claude models.
Since an update in August, Claude has been trained to end conversations involving persistently harmful or abusive user interactions. The revised policy goes further by expressly forbidding users from pursuing such exchanges. In its announcement, Anthropic said the rule is intended only for extreme cases in which users repeatedly act cruelly toward its models with no discernible purpose. The company clarified that it does not cover ordinary user frustration, pushback, dark creative themes, or model testing and research.
The update follows Anthropic’s collaborations with prominent religious scholars, which, according to reporting by The New York Times, in some cases involved efforts to persuade them that Claude could be conscious or have a soul.
Other sections of the policy prohibit broadly deceptive campaigns, including using Claude to operate fake accounts or fabricated news outlets. A dedicated section titled Do Not Undermine Democratic Processes sets out rules against deceiving voters or disrupting elections, strengthening the company’s safeguards around AI misuse in political contexts.
