GenAI Assessment and Language Equity: Drawing the Line between Support and Substitution

Generative artificial intelligence (GenAI) has changed higher education assessment by collapsing the practical boundary between language assistance and substantive authorship. For students who use English as an additional language (EAL), this collapse is not neutral. When integrity rules treat GenAI as a single category of unauthorised assistance, they risk converting linguistic disadvantage into integrity risk – the students most reliant on language support become the students most exposed to suspicion, inconsistent enforcement, and sanctions. This article argues that language equity in the GenAI era should be treated as a legality and academic ethics problem of assessment design. A defensible approach must separate legitimate language support (grammar, clarity, translation for comprehension, and surface-level editing) from impermissible substitution (generation or material restructuring of the reasoning, analysis, evidence, or sources the assessment intends to evaluate). Drawing on indirect discrimination logic, procedural fairness principles, and the academic integrity values of fairness and responsibility – and anchoring the analysis in an argument-based account of assessment validity – the article develops a model policy architecture and accompanying assessment design levers. The framework is purpose-based rather than tool-based, uses calibrated disclosure that is feasible for multilingual cohorts, and links enforcement to reasons, proportionality, and evidence standards that recognise the limits of detection-led reasoning. The contribution is a transferable governance blueprint for universities seeking to preserve standards while avoiding predictable, cohort-skewed unfairness in GenAI-era assessment.