News
Anthropic adds policy to protect Claude from abusive behavior
Anthropic updated its terms to prohibit sustained abusive or cruel interactions with its Claude model, warning that repeated cruelty may result in loss of access. The change reflects concerns about model distress and includes a Model Welfare team studying possible experiences of AI. The article ties this policy to the Hindu Ayudha Puja tradition of revering tools, urging users to treat AI respectfully for better outcomes, while noting mixed research on the impact of politeness on model performance.