<![CDATA[Article history: Accepted: 28 May 2026 Approved: 23 June 2026 The rapid integration of artificial intelligence (AI) tools in higher education has generated a paradox. While AI enhances accessibility and supports critical reading, it simultaneously risks fostering cognitive dependency that may erode students’ independent analytical skills. This literature review aims to identify the roles of AI for critical reading practices and the negative effects of overreliance on AI on critical reading skills. Using thematic synthesis based on qualitative content analysis, this study reviewed national and international publications from 2016 to 2026 sourced from Google Scholar, Scopus, and DOAJ. Findings reveal that AI functions as an accessibility enhancer, scaffolding tool, metacognitive catalyst, and self-regulated learning facilitator when integrated within structured pedagogical frameworks. However, excessive AI dependency is associated with cognitive offloading, epistemological passivity, atrophy of deep reading skills, and threats to academic integrity. Moderating variables, including prior reading habits, pedagogical context, and quality of AI interaction, shape the degree of risk. This review contributes a nuanced, evidence-based understanding of the AI–critical reading relationship, reframing AI as a conditional epistemological partner rather than an inherent threat or unconditional aid. Practical implications for instructional design and literacy policy in Indonesian higher education are discussed.]]>
