What Happens to Student Privacy When AI Enters the Classroom?

Abstract

Educational technology companies claim that artificial intelligence will improve learning. They also say AI will know each student as an individual learner and will make schools more innovative and efficient. These promises sound wonderful. However, we must look beyond these promises and ask questions about what is really happening in our classrooms when AI systems enter classrooms. AI enters schools with an insatiable appetite for data, including keystroke patterns, facial expressions, response times, learning trajectories, behavioural patterns, and, increasingly, biometric markers (Ismail, 2025). Schools are willingly feeding this appetite, often with minimal understanding of what happens to this data once it leaves the classroom. We have created a system where student privacy is not merely at risk. It has become a commodity in an educational technological complex that prioritises efficiency and profit overprotection. We teach students about digital citizenship and online safety while simultaneously enrolling them in learning platforms that harvest their most intimate intellectual and behavioural data. We ask students to trust institutions that have outsourced their trust to third-party AI vendors, many of which operate across borders with minimal oversight. In contexts like developing countries, where institutional capacity for data governance is already stretched thin, this vulnerability is magnified exponentially. The real controversy is not whether AI can improve education. It probably can, in limited ways. The controversy is that we are willing to sacrifice privacy for performance metrics we barely understand. We accept the premise that comprehensive surveillance of student learning is a necessary cost of educational innovation. We have normalised the collection of data that could be weaponised against students, not just today, but decades from now when algorithms we cannot predict make decisions about their opportunities based on digital footprints they left in third grade. Moreover, the conversation around consent in educational AI contexts is largely theatrical. Students do not consent. Parents sign forms they have not read. Administrators approve systems they do not comprehend. Teachers use tools they were not trained to evaluate critically. True consent requires a meaningful understanding of data flows, algorithmic decision-making, and long-term implications. These are luxuries rarely afforded in under-resourced educational systems. What makes this controversial is the silence.