<![CDATA[Generative artificial intelligence (GenAI) is increasingly embedded in English as a foreign language (EFL) learning, creating new forms of support as well as emotional uncertainty. Although learner emotions during GenAI use are increasingly documented, how teachers interpret and mediate these emotions remains underexplored, particularly in Indonesian EFL higher education. This study examines students' GenAI-related emotional experiences, lecturers' mediation practices, and points of convergence or divergence across the two perspectives. Employing a multi-perspectival interpretative phenomenological analysis, data were gathered through critical-incident interviews with three lecturers and six undergraduate students across three Indonesian universities. The analysis indicates that GenAI could shift rather than simply reduce anxiety: students described authenticity-related guilt, imposter-like feelings, frustration, and algorithmic helplessness when AI output exceeded their confidence or contradicted classroom knowledge. Lecturer accounts showed extrinsic regulation through permission-giving, demonstrations of AI fallibility, reassurance, and cognitive reframing. Cross-perspectival comparison further showed that these forms of support were weakened when institutional guidance on acceptable GenAI use was experienced as ambiguous. Responsible GenAI integration therefore depends on teacher-mediated emotional scaffolding, critical AI evaluation, and clear institutional guidance that supports learner agency and academic integrity.]]>
