Mental health AI is advancing quickly, but speed without the right foundation can introduce real risk.

Most AI tools are built on general-purpose models designed to be helpful and agreeable. In mental health, that approach can fall short. Effective care often requires challenging of unhelpful patterns, setting and reinforcement of boundaries, and knowing when to escalate.

In this session, Sword’s AI Research and Clinical leaders share how Mind was built differently, with a purpose-built model trained on clinically annotated data and evaluated through rigorous, clinician-led testing. You’ll leave with a clearer framework for what trustworthy mental health AI should look like and how to evaluate it.