Developing Critical Reading in Efl University Learners Through Notebooklm, a Source-Grounded AI Tool

Critical reading is a central skill for EFL university learners, who must evaluate claims, evidence and the author's stance in academic texts. At the same time, general-purpose generative AI tools make it easy to accept fluent answers without returning to the text, a habit described in the literature as cognitive offloading. This paper examines NotebookLM, an AI tool that generates answers only from the sources a teacher uploads, as a means of supporting critical reading in university EFL classes. It argues that source grounding turns verification into a visible, teachable step, and proposes a three-stage classroom sequence (THINK, SPEAK, CONNECT) in which learners form and discuss their own interpretations before consulting the tool. The paper is a pedagogical analysis based on published research; it does not report new experimental data.