Schools' Embrace of AI Connected to Increased Risks to Students (cdt.org)

A new report on AI use in the 2024–25 school year finds that as schools broaden and deepen AI adoption, student exposure to specific harms rises in tandem. The report flags four primary, correlated risks: cybersecurity incidents such as data breaches and ransomware targeting school systems; tech-enabled sexual harassment and bullying amplified by platforms and messaging tools; failures of AI systems that produce incorrect or harmful outputs; and troubling or unsafe interactions between students and AI (including non-consensual deepfake intimate imagery). It also highlights adjacent concerns around student activity monitoring, privacy risks for transgender and immigrant students, and gaps in AI literacy among educators and families. The significance for the AI/ML community is twofold: developers and deployers must account for adversarial and misuse scenarios (e.g., protecting training data, preventing deepfake generation, and hardening systems against ransomware) while schools and policymakers need practical safeguards. The report underscores the need for secure infrastructure, transparent model behavior, age-appropriate guardrails, robust privacy protections, and classroom AI literacy so that pedagogical benefits aren’t overshadowed by harms. Identifying concrete, tech-specific risks enables targeted prevention, incident response planning, and responsible procurement of AI tools for education.