UTSG 13 | Prof. Siobhan O’Flynn | Red Teaming GenAI Chatbots
May 2026
University of Toronto, ON, Canada
PI: Siobhan O'Flynn
The goal of this research project was to critically examine the impact of GenerativeAI Chatbots that are currently unregulated (or not rigorously regulated) in several parts of the world, including Canada. Specifically, focusing on the models that are conveniently available for use by adults and minors alike.
The five student researchers applied adversarial testing methods (red-teaming) to probe GenerativeAI Chatbots (LLMs) to test for any product liabilities due to their high risk anthropomorphic design, and examed their output in context of the current regulatory frameworks of such products in EU, US, and Canada.
The research poster above was presented at the Jackman Humanities Scholars-in-Residence 2026 Conference.
Two hundred copies of five unique zines depicting the findings from this research were also distributed at the conference.
All zines made for this project are archived at the University of Toronto Scarborough Library.