Understanding Human Behavior in the Age of AI
Behavioral AI Lab is an independent research and advisory initiative focused on the human layer of AI safety.
We help organizations understand how people interact with AI—and how behavioral science can reduce manipulation, strengthen trust, and build safer AI systems.
Current Initiative
Global Partnership for Detecting and Preventing AI-Enabled Manipulation and Scams
Behavioral AI Lab is developing an international initiative to address one of the fastest-growing forms of AI-enabled harm: manipulation and fraud conducted through personalized digital conversations.
The initiative builds on research analyzing more than 15,000 scammer messages from documented cryptocurrency romance scam cases across multiple countries and languages.
The goal is to advance open research, practical tools, and cross-sector collaboration that help protect vulnerable users before significant harm occurs.
Status: Proposal submitted to the United Nations AI Dialogue Partnerships Hub, 2026.
Focus Areas
Understanding Human Behavior in AI
AI safety depends not only on model performance but also on how people perceive, trust, and interact with AI systems. We study the psychological mechanisms that shape human-AI interaction and translate behavioral science into practical design principles.
AI-Enabled Manipulation & Scams
We analyze real-world datasets of AI-enabled scams and behavioral manipulation to identify early risk signals. Our research focuses on preventing harm before financial loss or psychological damage occurs through Safety by Design.
Safety by Design
Rather than reacting after harm occurs, we advocate designing AI systems that anticipate human vulnerabilities. We develop practical frameworks that help organizations build products that are safe, trustworthy, and aligned with human well-being.
Behavioral AI Research & Education
We bridge academic research and industry practice through publications, workshops, keynote talks, and open educational resources. Our goal is to help researchers, policymakers, and technology companies build more human-centered AI.
Founder
Behavioral AI Lab was founded by Tomomi Tanaka, PhD, a behavioral economist and AI safety practitioner whose career spans academia, international organizations, and the technology industry.
Tomomi began her career as an Assistant Professor at Arizona State University, where she researched behavioral economics and decision science. She later joined the World Bank as a Senior Economist, designing behavioral policy interventions and digital public sector initiatives across multiple countries.
She then moved into the technology industry, where she led Trust & Safety initiatives at Uber and Match Group, including establishing Match Group’s first dedicated Safety by Design team. She later led AI, automation, and enterprise technology programs at Amazon and Oracle, gaining firsthand experience in building AI-powered products and large-scale technology systems.
Across these roles, she recognized a common challenge: many of AI’s greatest risks stem not from the technology itself, but from how people interact with it. She also saw that behavioral science and Safety by Design remain underutilized in AI development, despite their essential role in shaping how people interact with AI and ensuring those interactions are safe, trustworthy, and beneficial.
This realization led her to found Behavioral AI Lab, bringing together behavioral science, AI safety, Trust & Safety, and Safety by Design to develop practical frameworks, conduct applied research, and help organizations build AI that is both innovative and safe for people.
Tomomi’s research has been published in leading academic journals, including the American Economic Review, and featured in the Harvard Business Review. She has presented on AI governance, Safety by Design, and human-centered AI at the United Nations Headquarters and other international conferences.
Tomomi Tanaka, PhD
Get in Touch
We welcome inquiries regarding research collaboration, consulting, speaking engagements, media interviews, and partnerships.