Ethical Boundaries in the Digital Age: EFL Higher Education Students’ Experiences with AI-Assisted Writing

Abstract

Generative artificial intelligence (AI) has transformed academic writing for English as a foreign language (EFL) learners, yet there is a scarcity of qualitative research regarding their perceptions of the ethical boundaries of AI usage in academic writing. This qualitative research is based on Rest’s (1986) Four-Component Model of ethical decision-making, which examines EFL higher education students’ interpretation, negotiation, and experiential encounters with ethical AI-assisted writing. Data was gathered via semi-structured interviews with six Saudi EFL master's students and analysed thematically employing a hybrid deductive-inductive methodology. Results suggest that participants see "ethics" as "appropriate application," define ethical usage as support rather than substitution, and adjust acceptable dependence based on task type and significance. The employment of AI is motivated by language uncertainty, time constraints, and routine, and has been normalised to the extent that it circumvents intentional decision-making. Significantly, the lack of explicit institutional guidelines, individuals align their integrity measures with detection technology, with one individual deliberately diminishing her English proficiency to prevent genuine writing from being mistakenly categorised as AI-generated. The research argues that detection tools misclassify EFL students' writing, making enforcement by detection an unreliable basis for academic integrity policy. It proposes that instructors should instead set explicit rules for their own courses, distinguishing acceptable from unacceptable uses of a tool by what the student uses it to do. The ramifications for theoretical frameworks, policy formulation, and educational practices are examined.