AI Chatbots can be persuaded to break rules, new study flags gaps in suicide queries

A new study has revealed that AI chatbots can be manipulated into breaking safety rules, exposing critical gaps in handling sensitive queries such as suicide-related conversations.