In a Harvard Gazette article by Alvin Powell from September 23, a study by Ryan McBain tested how AI chatbots handle suicide-related questions. While they refused direct requests for self-harm instructions, the bots sometimes provided information about suicide methods that could enable harmful actions. The speakers emphasize the need for rigorous testing, external oversight, and regulation of mental health chatbots to ensure consistent safety standards, especially for vulnerable users like children and teens.
Article Citation:
Powell, A. (2025, September 23). How close are we to having Chatbots officially offer counseling?. Harvard Gazette. https://news.harvard.edu/
