Explore
harms
A New Framework Puts Mental-Health Chatbots Through Adversarial Tests
The SIM-VAIL framework uses simulated vulnerable users, clinical review and automated scoring to examine how AI chatbots respond…
Social Media and Psychosis: What the Evidence Can—and Cannot—Tell Us
A review of 14 studies found no consistent link between social media use and worse psychotic symptoms, but…
Why Pro-Ana Communities Can Feel Supportive—and Still Carry Serious Risks
A systematic review of first-person accounts finds that pro-Ana communities may offer belonging and understanding while also reinforcing…