<![CDATA[The increasing use of artificial intelligence in higher education has raised concerns about misinformation, overdependence, and the weakening of learner agency. This qualitative case study explores how NotebookLM, as a source-grounded AI tool, supports learner autonomy in Basic Japanese learning among International Relations students at Universitas Jenderal Achmad Yani. The study was conducted in a Basic Japanese 1 course, where lecturer-curated materials were uploaded to the Learning Management System and used by students as the main knowledge base for NotebookLM-assisted learning. Data were collected through digital engagement observations, student reflective logs, and document analysis of Japanese curriculum vitae assignments. The findings show that NotebookLM supported students in verifying basic Japanese vocabulary, reviewing unfamiliar writing forms, revisiting course materials, and applying source-based expressions to personalized learning outputs. Evidence of learner autonomy appeared in students’ source consultation, independent vocabulary selection, self-monitoring, draft revision, and task-based decision-making. However, the study also found uneven patterns of AI use. Some students used NotebookLM critically as a learning support tool, while others tended to rely on it mainly for information retrieval. The findings suggest that NotebookLM does not automatically guarantee accuracy, eliminate AI hallucinations, or produce autonomous learners. Its pedagogical value depends on curated materials, clear task design, lecturer guidance, students’ digital literacy, and reflective learning practices. The study concludes that source-grounded AI can strengthen scaffolded autonomy when integrated responsibly into foreign language education, especially for beginner learners from non-language-major backgrounds. It offers a model for AI-assisted language learning that balances innovation, verification, and learner responsibility.]]>
