Grow Therapy and Stanford Partner on AI Safety Standards for Mental Health

Grow Therapy has entered a research partnership with Stanford University to develop evidence-based standards for the safe use of artificial intelligence in mental health care, announced in a press release.

The project is led by Dr. Jonathan Chen, Associate Professor of Medicine at Stanford, in collaboration with Stanford Psychiatry. The initial study will assess how leading AI models respond to mental health crises. It will also examine which design methods most effectively minimize harm and ensure appropriate handling of sensitive scenarios.

Stanford researchers will use de-identified clinical cases from Grow Therapy's AI coach, a conversational tool clients use between therapy sessions. A panel of psychiatrists will evaluate how often AI models produce harmful responses, test whether adding a safety review layer reduces risks, and assess other safety mechanisms from a clinical perspective.

Grow Therapy's AI coach includes multiple safety layers, automatic scoring for clinical quality and tone, and human review for flagged interactions. The company aims for the partnership to establish shared benchmarks for AI safety in mental health, drawing on real-world data to complement existing synthetic testing methods.