Canadian teachers' analyst: label K-12 AI 'high-risk' now

The op-ed at Tech Policy Press opens with the social-media parallel. Mia Travers-Hayward, a researcher and policy analyst at the Canadian Teachers' Federation, points to a New Mexico jury finding that "Meta knowingly deceived consumers about product safety, harming young people," a verdict that produced a $942 million judgment and binding design requirements. She also notes Meta's settlement with 47 states, the District of Columbia, and three territories for up to $17 billion over 10 years.

Her thesis is that the classroom is next. She cites a study of 26,811 Chinese students aged 12 to 18 that found AI users "performed better on homework, but worse on tests," and a randomized trial in which "high school students who used a standard GPT for math practice scored lower on their exams." A British Columbia survey of 4,423 educators reported that "86 percent worried about students' over-reliance on AI." A Common Sense Media poll of 1,017 American teens found 70 percent already use the tools for schoolwork.

The piece asks regulators to do four things: designate K-12 AI systems as high-risk within online safety legislation, ban design features such as "frictionless answers, engagement optimization, and anthropomorphic personas," require independent evaluation of any learning-outcome claim on the intended target demographic, and mandate child rights impact assessments before deployment.

Two AI analysts we track shared the essay the day it went up.