This study explored university EFL learners’ perceptions of AI-generated listening materials, emphasizing their instructional value, challenges, and effectiveness in enhancing listening skills. Conducted with 114 students in a Korean university EFL course, the research focused on three key questions regarding the opportunities, challenges, and effectiveness of these materials. Data were collected through a 12-item perception survey and open-ended responses during the final weeks of the semester. The quantitative results indicated generally positive attitudes toward the structured integration of passages, dialogs, and summaries. Students acknowledged that repeated exposure helped reinforce vocabulary learning and supported comprehension development. They particularly valued the adjustable audio playback speed and the option to revisit material, which fostered autonomous learning. However, significant challenges were identified, including unnatural intonation patterns, robotic voice delivery, and pronunciation inaccuracies. These issues were especially problematic for lower-proficiency students, who rely more on prosodic cues for comprehension. The findings suggest that AI-generated materials can significantly enhance EFL listening instruction when implemented with thoughtful pedagogical design, targeted guidance, and attention to the quality of synthetic speech. The study also discusses implications for EFL curriculum development and offers recommendations for future research.