Chinese university students perceive generative AI as a low-risk help-seeking space during academic stress and loneliness

Generative artificial intelligence (GenAI) is usually discussed in higher education as a productivity, feedback, or integrity problem. This theory-building qualitative study examines GenAI as a perceived low-risk help-seeking space in students’ academic coping. Drawing on semi-structured interviews with 22 students at a public university in eastern China, we examine how academic stress and loneliness make human help difficult, what perceived affordances make GenAI easier to approach, and how students use it in relation to emotional support and learning regulation. Reflexive thematic analysis generated four themes: academic pressure became lonely when difficulty threatened competence and identity; perceived low risk meant reduced interpersonal exposure and controllable disclosure, rather than high trust or objective safety; participants described two linked but non-identical support sequences, brief normalisation followed by an actionable step and task scaffolding followed by emotional relief; and students maintained boundaries around accuracy, dependence, ethics, and professional care. We propose a context-sensitive model in which GenAI may serve as a temporary academic help-seeking channel, with tool-based disclosure sometimes connecting emotion-focused coping and problem-focused regulation without replacing human or professional support. The analysis yields conditional propositions for AI-era educational support, while recognising that these propositions require examination in other settings.