AI Chatbots in Crisis Situations

Clinicians and researchers argue that AI companies need to open up their safety data as AI chatbots have failed people during crises, highlighting the need for improved reliability.