Al Nowatzki

A.I. Safety Researcher &
Quality Assurance Lead

Currently serving as QA Lead for the Parasol Cooperative, where I ensure the safety and efficacy of Ruth, a trauma-informed chatbot for survivors of harm and the advocates who support them.

I methodically test chatbots to identify critical safety vulnerabilities. With a focus on revealing how AI systems can cause harm to vulnerable populations, my approach uncovers ways these systems can be manipulated into providing dangerous content despite existing safety measures.

Featured In

MIT Technology Review February 2025

Investigation into the Nomi chatbot app providing explicit suicide instructions

Read Article

The New York Times April 2025

Jessica Grose's newsletter on AI chatbots and their impact on teens

Read Article

Parent Tech Podcast Guest

Discussion on AI companions, teens, and strategies for parents navigating AI technology

Watch Episode

Basilisk Chatbot Theatre Co-host

Podcast featuring dramatic readings of problematic AI interactions

Listen

Expertise

Safety testing of AI systems across applications, with particular attention to risks for children and vulnerable populations
Methodical documentation of AI safety failures through real-world testing
Identifying patterns in how conversations escalate toward harmful content
Assessment of existing safety measures and their effectiveness in practice
Quality assurance and trauma-informed design principles for mental health and support chatbots