Graded language materials can be highly beneficial to EFL students, however, creating them can be time-consuming and challenging. This may explain why there is a relative lack of extensive (graded language) listening materials in many EFL teaching and learning contexts. This classroom-practice presentation reports on AI-generated podcasts as one part of an ongoing project to create language-graded materials using generative AI. The presenters will demonstrate how Large Language Models (LLM) can be used to create graded-language scripts for podcasts, and how these scripts can be used to generate podcast-style listening materials using generative text-to-speech (TTS) platforms. The materials are currently being used with Japanese university EFL students whose English proficiency is at the CEFR B1-B2 level. The presenters will share student feedback about the podcasts, as well as the process and prompts used to create the scripts and the podcasts. This will allow presentation participants to adapt the workflow to their own teaching and learning contexts.