From Promise to Practice: A Guide to the Responsible Integration of Generative AI in Large-Scale Educational Assessment under the EU AI Act

This white paper addresses the responsible integration of generative artificial intelligence (particularly large language models, LLMs) into large-scale assessments (LSA) in education against the backdrop of
the European Union’s Artificial Intelligence Act (AIA). The paper aims to address uncertainties among experts about how to integrate AI tools efficiently without compromising legal, ethical, or psychometric
standards. We seek to provide a clear framework (including checklists and examples) to guide compliance efforts during the current transitional phase, when official technical standards for operational implementation of the AIA are still largely lacking. The strategic imperative should be to proactively create compliance structures to maintain the balance between technological progress and the protection of fundamental rights. For this purpose, informed and professional human judgement remains indispensable in strengthening public confidence in the validity and fairness of AI-supported educational assessment methods.