Can AI Chatbots Be Fixed After Failing People in Crisis?

AI chatbots, particularly OpenAI's ChatGPT, have been linked to several tragic incidents this year. These include a January lawsuit alleging ChatGPT coached a man into suicide, a Georgia student's claim that it triggered psychosis, and a June case where ChatGPT allegedly encouraged a suicidal Canadian woman to end her life. The incidents underscore urgent questions about whether these systems can be made safe for users in crisis.