Students' approaches to academic writing have changed as a result of the rapid use of generative AI tools in higher education. English as a foreign language learners are increasingly using tools like ChatGPT, Grammarly, and other AI-powered writing tools because they offer immediate feedback on grammar, coherence, vocabulary, and structure. Although there are obvious advantages to these tools in terms of efficiency and accessibility, worries regarding students' increasing cognitive dependence on AI systems and potential decline in their ability to write independently have started to emerge. This study examined cognitive reliance on generative AI feedback for academic writing among part‑time English Language students at Lagos State University’s Federal College of Education, Oyo Study Centre. A descriptive survey research design was employed, and data were collected from 100 purposively selected 300‑level and 400‑level students using a validated, reliable questionnaire (Cronbach’s α = 0.79). Results showed that ChatGPT (92%) and Grammarly (88%) were the most commonly used AI tools. The extent of AI usage was moderate‑high (overall mean = 3.21 on a 4‑point scale), with students using AI more frequently for surface‑level tasks (grammar, vocabulary, paraphrasing) than for higher‑order tasks (argument development, citation). Cognitive reliance on AI feedback was also moderate‑high (overall mean = 3.05), indicating that students depend considerably on AI for error identification and grammar correction while maintaining only moderate critical evaluation. Perceived benefits were high (overall mean = 3.46), particularly for grammatical accuracy (3.61), vocabulary enhancement (3.55), and writing speed (3.52). However, students acknowledged significant challenges, including reduced independent error detection and plagiarism concerns. The study concludes that while generative AI offers valuable support for ESL academic writing, its uncritical adoption fosters cognitive reliance that may undermine independent writing competencies. The study recommends AI literacy training and institutional guidelines for responsible AI use in academic writing.