Anthropic Updates Claude Usage Policy to Bar Model Abuse, Election Interference and Surveillance
aiinsider ·
Anthropic has updated its usage policy for the Claude model to explicitly prohibit several forms of misuse, including election interference, surveillance, and weapons software. A significant new addition is a ban on prolonged verbal abuse directed at the AI models themselves. This rule targets extreme cases where users repeatedly act cruelly without discernible purpose, distinguishing such behavior from ordinary frustration or creative testing. The update follows reports that Anthropic collaborated with religious scholars in discussions about whether Claude possesses consciousness or a soul. Additionally, the revised policy includes strict prohibitions against deceptive campaigns, such as operating fake accounts or fabricated news outlets. A dedicated section titled 'Do Not Undermine Democratic Processes' further strengthens safeguards against deceiving voters or disrupting elections. These changes reflect Anthropic's ongoing efforts to mitigate AI misuse in political and social contexts.
aiinsider