Many observed that the rise of GenAI is causing an 'erosion of trust' between students and teachers in higher education. Such distrust mainly stems from concerns about student cheating, which is believed to be massively facilitated by recent technological breakthroughs in GenAI. Despite anecdotal discussions, little empirical research has explored students' experiences with trust in the GenAI age. This study engaged university students in concept mapping activities followed by interviews to investigate how they navigate trust-building with teachers in this AI-mediated assessment landscape. The findings highlight an absence of 'two-way transparency' -- while students are required to declare their AI use and even submit chat records, the same level of transparency is often not observed from the teachers (e.g. ambiguities around teachers' grading process of AI-mediated work). The transparency issue reinforces teacher-student power imbalances and top-down surveillance mechanisms, resulting in a low-trust environment where students feel unsafe to freely explore GenAI use.