<p>This duoethnographic study explores how two university language teachers in China and Japan understood, negotiated, and attempted to cultivate shared epistemic agency in AI-enhanced classrooms. In this study, shared epistemic agency refers to the collaborative capacity to engage in knowledge creation through reflection, judgment, and responsibility in relation to AI-mediated pedagogical practice. Drawing on co-constructed reflexive journals, sustained Zoom and in-person dialogues, and contextual prompts from student artefacts, classroom notes, and selected AI interaction logs, the study traces how the two teachers interpreted the conditions and tensions involved in fostering such agency. Four recurring concerns became central to their dialogue: reflexive task design, recursive reflection on AI use, learner ownership and responsibility, and contextual challenges and strategies in cultivating shared epistemic agency. Their accounts show how local ideologies, language norms, institutional policies, and cultural expectations may shape what counts as responsible and meaningful AI use in each setting. The study highlights the value of duoethnographic reflection for illuminating the situated conditions, tensions, and pedagogical strategies through which shared epistemic agency may be cultivated. It also underscores the need to foreground learners’ epistemic participation in AI-era language education while critically attending to the cultural and institutional conditions that mediate human-AI collaboration.&#xa0;</p>