Assistant or Ghostwriter? EAP Instructors’ Views on the Opportunities and Risks of GenAI in L2 Writing

Generative artificial intelligence (GenAI) systems are able to generate, revise and comment on second language (L2) student writing. However, less is known about where L2 writing teachers locate the opportunities and risks of such systems across the writing process. To that end, this study examined how English for academic purposes (EAP) writing instructors at two universities in Oman position GenAI within the L2 writing process. It focused on, first, the stages at which instructors locate GenAI's opportunities and risks, second, their views on GenAI as a source of feedback, and lastly their preparedness to manage academic integrity in AI-assisted L2 writing. A cross-sectional questionnaire combining Likert-scale items and open-ended prompts was completed by 51 EAP instructors , of whom 49 were retained for analysis. Closed items were analysed with descriptive statistics, Spearman correlations, McNemar tests and Mann–Whitney U tests. Open-ended responses were coded by writing stage through qualitative content analysis. The results showed that L2 writing instructors located GenAI's opportunities mainly in their own preparation of materials and model texts and in feedback. On the other hand, they highlighted its risks in students' drafting and in assessment. Revising and editing was contested, cited equally often as an opportunity and a risk. EAP instructors were more confident in using GenAI to prepare materials than in designing students' use of it at each writing stage. Just over half agreed that AI improves their feedback. This view was related to their overall attitude towards GenAI but not to their confidence as feedback providers. Most instructors felt responsible for checking students' GenAI use and confident in setting boundaries, but their confidence in maintaining academic integrity was unrelated to this sense of responsibility. More EAP instructors reported knowing how to use AI detectors than understanding how such detectors work. The study concludes that EAP instructors' readiness to integrate GenAI into L2 writing teaching and learning is uneven across the L2 writing process. Professional learning and development should therefore target the pedagogical design of student-facing AI use, the evaluation of AI feedback, AI detection literacy, and L2 writing assessment redesign.