Oxford and DfE Launch Six-Month Study: Can Student Data Power AI Without Compromising Children's Rights?

Pupil Data as a Public Good

The University of Oxford has begun a six-month research project with the UK Department for Education (DfE) on one of the most sensitive questions in education technology: how data generated by school students could be used in AI research without compromising children's privacy or digital rights.

The project — titled Publicly Funded Pupil Data as a Strategic Asset and Public Good — is funded through an Economic and Social Research Council (ESRC) React Award and co-designed with the DfE. It brings together researchers in education, law, AI governance and learning science, with contributions from Ofqual, Oak National Academy, and expert support from Baroness Beeban Kidron, the UK's most prominent children's digital-rights advocate.

What the Study Will Produce

  • A proposed governance framework for student-generated data.
  • Guidance on implementing a Pupil Data Commons — a model for how student-generated information could be managed and accessed responsibly (explicitly not yet an operational data-sharing service).
  • Recommendations for future AI assurance and certification processes for educational technologies.

“Together we hope to prepare a set of research-informed recommendations as to how England can best safeguard student data and the digital rights of our young people at schools across the country,” said Dr. Sara Ratner, who leads the project at Oxford's Department of Education.

Oak National Academy Contributes Real Data

Oak National Academy will contribute fully anonymized datasets of multiple-choice responses — giving the research team real student-response data to test governance approaches against. Oak already makes its curriculum content and code openly available for developing educational tools. “Robust data is essential for AI to compute how children learn, and in turn, improve their outcomes. But responsible handling is non-negotiable: pupil privacy cannot be compromised,” said CEO John Roberts.

Ofqual adds the regulatory lens. “Ensuring student data is used responsibly in the development and use of AI tools is critical,” said Jo Handford, Ofqual's Director of Assessment Integrity & Innovation.

Why It Matters Beyond England

The study lands at a pivotal moment. AI vendors increasingly want real educational data to train and evaluate their models, while privacy advocates warn that classroom data was collected for teaching, not for product development. If England produces a workable model — with anonymization standards, governance rules and certification requirements — it could become the template for other education systems facing the same tension. The findings, due next year, are intended to inform policy, responsible AI development and future assurance requirements for EdTech sold into schools.

También disponible en: English Svenska Deutsch