生成式人工智能时代下的青少年隐私保护

Safeguarding youth privacy in the age of generative artificialintelligence Tiffany Kwok and Christelle Tessono | March 2025 Acknowledgements The Dais is a public policy and leadership think tankat Toronto Metropolitan University, working at theintersection of technology, education and democracyto build shared prosperity and citizenship for Canada. This project has been funded by the Office of thePrivacy Commissioner of Canada (OPC); the viewsexpressed herein are those of the author(s) and donot necessarily reflect those of the OPC. For more information, visitdais.ca20 Dundas St. W, Suite 921, Toronto, ON M5G 2C2 Design and IllustrationMariana Rodrigues @daisTMU Copy Editor Suzanne Bowness, CodeWord Communications How to Cite this Report Contributors Nina Rafeek DowAndré Côté Kwok, Tiffany and Christelle Tessono.(Gen)eration AI: Safeguarding youth privacy in the ageof generative artificial intelligence. The Dais. 2025.https://dais.ca Acknowledgments: This project could not have been made possiblewithout the participation and insights provided by allthe interviewees: © 2025, Toronto Metropolitan University350 Victoria St, Toronto, ON M5B 2K3 Howie Bender, Rubicon PublishingPaulla Bennett, York Region District School BoardAndrew Bieronski, QuizizzJovine Chan, York Region District School BoardAngela Chen, Stanford Accelerator for LearningTom D’Amico, Ottawa Catholic School BoardRemo George Joseph, QuizizzIndra Kubicek, Digital MomentJacques Marcoux, Canadian Centre for ChildProtection (C3P)Brenda McPhail, McMaster UniversityDavid Porter, David Porter and AssociatesValerie Steeves, University of Ottawa This work is licensed under a Creative CommonsAttribution-NonCommercial-ShareAlike 4.0International License. You are free to share, copyand redistribute this material provided you: giveappropriate credit; do not use the material forcommercial purposes; do not apply legal terms ortechnological measures that legally restrict othersfrom doing anything the license permits; and if youremix, transform, or build upon the material, you mustdistribute your contributions under the same license,indicate if changes were made, and not suggest thelicensor endorses you or your use. The Dais proudly engages a diverse group of fundersto support and catalyze our work, consistent with ourvalues, and subject to a thorough internal review. As anon-partisan, public-interest institute, we only acceptfunds from organizations that support our missionand enable us to undertake work independently, withfull editorial control. The names of all of our financialsupporters are publicly and transparently displayedon all online and printed material for each project orinitiative. Authors Tiffany KwokPolicy Analyst Tiffany Kwok (she/her) is a policy analyst at the Dais. She ispassionate about tech and urban policy, service design, andresearch, and has experience working in the public sector andvarious non-profit organizations both in Canada and the UK.Tiffany holds a Bachelor’s degree in Political Science and UrbanStudies from the University of Toronto and a Master’s degree inDigital Technologies and Policy from University College London. Christelle Tessono Policy and Research Assistant Christelle Tessono (she/her) conducts research at theintersections of digital technology, human rights, andgovernance. This has led her to work on a variety of projectsrelated to political advertising on social media platforms, gigwork, facial recognition technology, privacy, and AI governance.Christelle holds a Bachelor of Arts in political science from McGillUniversity and is currently pursuing graduate studies at theUniversity of Toronto’s Faculty of Information. Table of Contents 5Executive Summary 8Introduction 10What is Generative AI? 10The whole wide world of generative AI12GenAI and privacy harms14Harms beyond privacy 15The Policy Landscape: GenAI Privacyand Data Protections for Children 15Canadian policy landscape17Zooming in: provinces and school boards18Lessons from other jurisdictions 22Best Practices for SafeguardingChildren and Teen Privacy 22Privacy design principles as a common foundation 23For policymakers: Youth interests as a priority for genAIgovernance 23For technologists and privacy practitioners: Privacyconsiderations in the development and maintenance ofgenAI tools 25For educators and parents: Protecting youth privacy intheir exploration and use of genAI tools 28Conclusion Executive Summary BOLD IDEA:As generative AI tools become commonplace for children and teenagers, so mustgovernments, school administrators, and technologists introduce new practical privacy policy interventions. Chatbots as digital companions, AI-driven teachingtools, and content-generation tools are all examplesof how generative AI (genAI) is transforminghow young people learn, connect, and expressthemselves. This report examines the unique privacy riskschildren and teenagers face when using genAI tools,particularly in educational