Materials development in English Language Teaching (ELT) has traditionally been informed by research in Second Language Acquisition and Applied Linguistics, which identifies authenticity, meaning, communication, scaffolding, and collaboration as core features of effective materials (Tomlinson, 2011). Generative AI complicates these principles by challenging what counts as effective materials and by raising new questions about how materials are selected, developed, and adapted in context. It can therefore be framed either as a disruptor of existing practice or as a catalyst for pedagogical integration (Moorhouse et al., 2024; Law, 2024; Barrot, 2023; Byrne, 2025).

Research on generative AI chatbots in writing pedagogy highlights affordances such as support for process writing, personalised guidance, and detailed feedback, but also raises concerns about biased or inaccurate outputs, plagiarism, and overreliance on AI systems (Ahn et al., 2024; Pack & Maloney, 2023; Li et al., 2024; Creely et al., 2025). While prompt engineering may improve output quality (Bender, 2024; Dornburg & Davin, 2024; Wünsch-Nagy, 2025), less attention has been paid to teacher cognition: teachers’ beliefs and knowledge about language teaching, which shape how they interpret, evaluate, and adapt AI-generated materials (Borg, 2003).

This study investigates upper-secondary English teachers’ use of ChatGPT for developing writing materials and how this interaction influences their pedagogical thinking. Data from reflective workshops, semi-structured interviews, and ChatGPT interaction scripts are analysed thematically. Emerging findings suggest that ChatGPT can support materials design while also prompting reflection on teachers’ beliefs, agency, and situated decision-making in responsible AI integration for language education in contemporary classrooms.