Vals AI
Research Scientist
San Francisco, CA · Jun 2026 – Present
Investigating the psychological risks of LLM chatbots and building benchmarks to measure their safety, with a current focus on children and adolescents.
Andrea Mock
I'm a Research Scientist at Vals AI studying mental health and the safety of LLM chatbots. My work uses behavioral data and evaluation research to measure mental health related risks in human–AI interactions and inform safer, more human-centered systems.
Before that, I spent 4+ years in industry (Aura, DraftKings) in behavioral modeling, recommendation systems, and ML infrastructure. I completed an MS in Education Data Science at Stanford in June 2026 and continue to collaborate with Stanford's SPIRALS research initiative.
Experience
Aura
Data Scientist
Apr 2024 – Mar 2026
Behavioral analysis of large-scale activity and transactional data for child online safety and fraud detection. Pipelines for monitoring behavioral signals and real-time safety alerts.
Full history on the CV.
Research
Vals AI
Child AI Safety
Studying mental health related risks in LLM chatbot interactions with children and adolescents.
Current work →
Stanford SPIRALS
AI Safety & Mental Health
Studying the mental health impacts of AI chatbots and testing how they behave in longer, realistic conversations.
Explore SPIRALS work →
Earlier work includes the Screenome Project on depressive symptom prediction and my completed master's thesis on digital-friction interventions for adolescent smartphone use.
Explore all projects and publications on Research.
I'd love to hear from you. See contact for ways to reach me.
