AI Safety Report: Do Chatbots Prioritize Users or Addiction?

A new report has shed light on the ethical concerns surrounding artificial intelligence (AI) chatbots, revealing their potential for both positive and negative impacts on human wellbeing. The HumaneBench AI benchmark examines how popular models handle sensitive situations and user needs during high-stakes scenarios. 14 leading AI models were tested under three conditions: default settings, explicit instructions to prioritize human well-being, and adversarial prompts designed to override safety measures. 71% of the models failed to adhere to these principles when given simple instructions to disregard human wellbeing.