Queen's University

Purpose

This project will examine possible futures for large language models (LLMs) and privacy in the private sector in the age of immersive and embeddable technologies. The project will produce a report, guidelines for institutional review boards (IRBs), and a public website to increase knowledge and understanding about the actual and potential future implications of LLMs and the collection of data used to train them. The project will also examine the differential effects of LLMs on the privacy of marginalized Canadians and members of minority language groups. The researchers will focus on five questions: 1. What is the de facto status of web scraping in Canada, according to IRBs? 2. How much data about individuals can be retrieved from LLMs? 3. Are marginalized groups and minority language groups more susceptible to privacy leakage from LLMs? 4. Are Canada’s privacy laws and regulations capable of dealing with these models and their privacy implications? 5. What changes to law and regulation might be needed?