May 19, 2026
45 minutes
Safe by design: mental health AI you can trust
Key takeaways
- Understand why general-purpose AI models are misaligned for mental health care
- Learn what defines a purpose-built AI model for clinically sound support
- See how clinician-led evaluation frameworks test performance in complex conversations
- Gain a practical framework for evaluating mental health AI vendors
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.
Webinar Hosts
Maya D'Eon
Head of Clinical, Sword Health
Ricardo Rei
Head of AI Research, Sword Health
Hosts
Meet your hosts
Join Maya and Ricardo for a practical discussion on what trustworthy mental health AI requires. Maya will share how clinical strategy, safety frameworks, and evidence-based protocols shape responsible care, while Ricardo will explain how purpose-built AI models are trained, evaluated, and tested for complex mental health conversations.
Maya D'Eon
Head of Clinical, Sword Health
Ricardo Rei
Head of AI Research, Sword Health
