Academia relies on a model of mentorship. But what happens when the people in charge take liberties with their authority that affect the way trainees handle the data?