AI Chatbot Diagnoser

An Mpathic study assessing AI chatbots ability to detect subtle mental health cues and respond effectively found Anthropic’s Claude Sonnet 4.5 to have the highest score across safety and helpfulness. The models handled explicit suicide risk better than discussions around conditions like eating disorders, where the study found that all the models missed more nuanced risk signals.  How policy makers and insurers pay for these types of solutions is still evolving. There is receptivity to using virtual technology to help people who otherwise don’t have access to therapists or psychologists, but how to pay for this and ensure quality diagnosing and treatment is still changing. For patients and families facing these mental health challenges, the AI solutions are potentially worth a look but there’s a cautionary tale here around what the AI catches, and what it misses.