Ars Technica AI Security & Safety · Ago 7, 2026 13:49 imp:65 AI chatbots have failed people in crisis. Can that be fixed? Clinicians and researchers say AI companies need to open up their safety data. Leer el original en Ars Technica AI →