In an essay for MindSite News, Tom Insel, the psychiatrist and former director of the National Institute of Mental Health, weighs the evidence on whether AI chatbots are safe and effective for mental health. He draws a sharp line between general-purpose models and tools built specifically for the task, likening a general model asked to act as a therapist to a Swiss Army knife asked to perform surgery. Citing recent evaluations of Ash, Slingshot AI's purpose-built system, he notes it produced harmful responses in roughly 6% of benchmark cases compared with 42% for a leading general-purpose model, and that an analysis of 20,000 real conversations found a 0.38% false-negative rate for suicide and self-harm. Insel concludes that the real question is not whether people will use chatbots but whether we build them wisely, with clinical safeguards and warm handoffs to human care.