“We cannot incorporate them because …”: Challenges of integrating generative AI into online formative assessments in EFL contexts

While some prior studies have explored the challenges associated with integrating AI into education, few have undertaken a comprehensive investigation into the barriers to incorporating Generative AI into online formative assessment, especially within English as a Foreign Language (EFL) context. This mixed methods study aims to explore the challenges encountered by EFL teachers in Iran while integrating Generative AI into online formative assessments. A purposive sample of 30 Iranian EFL teachers participated in the study, providing valuable insights. Qualitative data was gathered through focus group discussions, with teachers categorized into three groups based on educational background and teaching context: school, university, and institute. The findings of this study shed light on three primary themes of challenges encountered by the EFL teachers: technical challenges, operational challenges, and ethical and equity-related challenges. Furthermore, in the quantitative phase of the study, a ranking scale was employed to gauge the perceived significance of these challenges. The implications of this research extend to EFL teachers, AI tools developers, policymakers, and teacher trainers. It guides educational systems in identifying support needs, such as infrastructure and training, and informs policy reforms to promote ethical AI integration, enhancing teaching practices and student outcomes in EFL education. • This study examines EFL teachers’ challenges in integrating generative AI into online formative assessment. • I identified technical, operational, and ethical–equity barriers shaping teachers’ AI use. • Teachers viewed technical issues—such as low AI literacy and poor infrastructure—as the most significant obstacles. • The findings offer human-centered recommendations to support more usable, equitable AI-based assessment systems.