Anthropic prohibits users from subjecting Claude to persistent and unnecessary abuse or cruelty

Anthropic updated its user policy, prohibiting users from exhibiting persistent and gratuitous abuse or cruelty toward its models. The policy does not apply to common user frustration, model testing, or dark creative themes. Anthropic had previously allowed its large language models to end conversations when users were persistently harmful, and when it launched that feature in August last year, it positioned it as a safeguard for AI welfare, stating that the potential moral status of Claude and other large language models remains highly uncertain. Regarding whether AI has consciousness, Anthropic CEO Dario Amodei said this possibility cannot be ruled out, while OpenAI CEO Sam Altman said that imposing religious-style authority or abandoning human judgment on AI models is a real safety problem.