Are A.I. Therapy Chatbots Safe to Use? (www.nytimes.com)

After a harrowing personal episode, Brittany Bucicchia turned to an AI therapy chatbot called Ash and found it helpful: the system remembered past conversations, suggested topics, summarized sessions and even provided a crisis telephone number when she described suicidal thoughts. Her experience illustrates a fast-growing wave of automated mental‑health tools — built by startups and academics — that aim to expand access to psychological support by offering conversational, memory‑aware, triage-capable interactions outside traditional therapy. That growth has prompted regulators to act: the Food and Drug Administration held its first public hearing to consider whether AI therapy chatbots should be treated as medical devices. The stakes are technical and ethical. On the plus side, chatbots can scale care, provide continuity, and perform automated crisis prompts or referrals. On the minus side, they vary widely in clinical validation, risk‑detection accuracy, data handling and transparency; failures can mean missed crises, inappropriate reassurance or privacy breaches. Classifying these systems as medical devices would impose evidence standards, premarket review and post‑market surveillance — potentially raising safety and accountability but also shaping innovation and access in digital mental health.