Abstarct: This study investigates the impact of ChatGPT-baxsed feedback on the writing proficiency of Iranian EFL intermediate learners, comparing its effectiveness with traditional teacher feedback. Utilizing a quasi-experimental design, 32 learners from Safir English Institute in Qom, aged 18–25, were selected via convenience sampling and divided into an experimental group (n=17, receiving ChatGPT feedback) and a control group (n=15, receiving teacher feedback). Both groups completed a pretest and posttest, writing 150–200-word paragraphs baxsed on IELTS prompts, scored using Hyland’s (2003) rubric for content, organization, and language accuracy. Over eight sessions, the experimental group received AI-generated feedback via ChatGPT, while the control group received direct teacher feedback on grammar, vocabulary, and structure. Statistical analysis using paired and independent samples t-tests revealed significant improvements in both groups, with the ChatGPT group outperforming the control group, particularly in organization and coherence. These findings align with Vygotsky’s sociocultural theory and task-baxsed language teaching, highlighting ChatGPT’s potential to provide timely, scalable feedback that enhances writing skills. The study suggests integrating AI and teacher feedback to optimize EFL writing instruction, addressing challenges like teacher workload and delayed feedback, though limitations include a modest sample size and short intervention period. Future research should explore longitudinal effects and learner perceptions to further refine AI applications in language education.