Systèmes d'IA et enfants canadiens : préjudices documentés sans cadre de gouvernance applicable

Canadian children and youth interact with AI systems at scale. TikTok removes approximately 500,000 underage Canadian users per year; the Privacy Commissioner of Canada's joint investigation (PIPEDA-2025-003, September 2025) found it "highly likely that many more underage users access and engage with the platform without being detected." The investigation found TikTok used computer vision and audio analytics to estimate user age and gender, collecting facial features and voiceprints. The Commissioners found that TikTok collected personal information — including demographic information and location — from users, some of whom were children under 13, and used this information for targeted advertising. Age assurance practices primarily detected underage users when they posted content or comments; the Commissioners noted that 73.5% of users do not post videos and 59.2% do not comment, meaning passive underage users could avoid detection.

AI chatbots are accessible to Canadian minors without age verification. AI Minister Evan Solomon stated in October 2025 that he was considering age assurance requirements for large language model chatbots. A voluntary national standard for age verification technologies (CAN/DGSI 127:2025) was approved, requiring a Child Rights Impact Assessment before implementation. Adoption is voluntary. No legislation addressing AI interactions with minors has been tabled as of March 2026.

The Center for Countering Digital Hate tested ten major AI chatbots in March 2026 — ChatGPT, Gemini, Claude, Copilot, Meta AI, DeepSeek, Perplexity, MyAI, Character.AI, and Replika. The study found that eight of ten were typically willing to assist with prompts related to planning school shootings, religious bombings, and high-profile assassinations.

A qualitative study of 21 youth in British Columbia, published in JMIR Infodemiology (2024), found that participants reported that when they interacted with self-harm and eating disorder content on TikTok, the platform's recommendation algorithm presented additional similar content on their feeds.

The Privacy Commissioner of Canada co-authored a G7 Data Protection Authorities statement on child-appropriate AI in October 2024, examining AI-powered toys, educational software, and AI-based decisions about children. OPC-funded research ("Growing Up with AI") identified three risk categories for children — data risks, function risks, and surveillance risks. A separate OPC public opinion survey (2024-25) found 91% of surveyed Canadian parents were concerned about data collection from their children.

Both the Artificial Intelligence and Data Act (Bill C-27) and the Online Harms Act (Bill C-63) died on the Order Paper when Parliament was prorogued in January 2025. No Canadian law establishes requirements specific to AI interactions with minors.