This study examines the pedagogical use of artificial intelligence (AI)–generated written corrective feedback (WCF) in a first-year university writing course at a private university in Japan. Conducted over a 14-week term with three classes of first-year students (n = 62), the study implemented an eight-week intervention in which learners completed three-minute speed writing tasks followed by AI-generated WCF provided by ChatGPT to support self-correction and revision. Using a mixed-methods design, quantitative data were drawn from analyses of learner writing samples, focusing on error types, frequency, and AI-generated corrections. Qualitative data were collected through a post-intervention survey examining learners’ perceptions of the usefulness of AI-generated WCF, their self-reported proficiency using AI tools, and attitudes toward AI integration in writing courses. The study addressed two research questions: (1) Did AI-generated WCF raise learners’ language awareness in writing? and (2) What are learners’ perceptions of AI use in university writing courses? Findings indicate that repeated interaction with AI-generated feedback can enhance learners’ noticing of linguistic form and accuracy, positioning AI as a language awareness raising tool rather than solely an error corrector. However, effectiveness depends on the development of learner AI literacy, particularly for autonomous use inside and outside the classroom.