NEWS

In its second session, IFH Club continued its exploration of safe and trustworthy AI in healthcare by focusing on approaches for improving safety in mental health applications. The discussion centered on domain-specific constitutional AI and related work on helping conversational systems respond more safely and supportively during sensitive mental health interactions. A central theme of the session was the importance of designing AI systems that do more than simply stop when risk is detected. Participants discussed how AI tools in mental health settings should be developed to respond with care, responsibility, and clinical awareness while remaining aligned with appropriate safety principles.