Artificial Intelligence (AI) is rapidly changing higher education, offering new opportunities to support diverse learners, including students with dyslexia, while raising important questions about ethics and assessment. Existing work on Universal Design for Learning (UDL) and assistive technologies highlights persistent barriers in reading, writing and planning, but there is limited guidance on how generative AI (GenAI) can be integrated into academic work safely and accessibly. This study addresses that gap through a convergent parallel mixed-methods design with qualitative priority (QUAL + quan). Semi-structured interviews with specialist Student Support and Wellbeing Services (SSWS) were conducted alongside an online student survey to identify recurring challenges, current tool use and attitudes towards AI. The analysis reveals difficulties with academic writing, critical reading and time management, patchy and sometimes unsustainable adoption of assistive tools and AI, and uncertainty about acceptable AI use and its implications for academic integrity. In response, the paper proposes a dual-component, UDL-aligned system for responsible AI-supported assessment. The instructor-facing AI Engagement Spectrum helps academics set clear, assignment-specific rules for permissible AI use, grounded in constructive alignment. The student-facing AI Scaffold translates these rules into a structured workflow that guides students with dyslexia through reading, planning, drafting and revising with AI support while preserving authorship and critical engagement. The primary contribution is a transferable framework that links institutional policy, pedagogical intent and students’ study practices. It offers SSWSs, lecturers and educational designers a practical starting point for implementing AI in ways that are accessible, transparent and focused on supporting students with dyslexia.
Ram et al. (2026) studied this question.