Protection of student privacy is critical for scaling up the use of learning analytics (LA) in education. Poorly implemented frameworks for privacy protection may negatively impact LA outcomes and undermine trust in the discipline. To design and implement models and tools for privacy protection, we need to understand privacy itself. To develop better understanding and build ground for developing tools and models for privacy protection, this paper examines how privacy hitherto has been defined by LA scholars, and how those definitions relate to the established approaches to define privacy. We conducted a scoping review of 59 articles focused on privacy in LA. In most of these studies (74%), privacy was not defined at all; 6% defined privacy as a right, 11% as a state, 15% as control, and 16% used other approaches to explain privacy in LA. The results suggest a need to define privacy in LA to be able to enact a responsible approach to the use of student data for analysis and decision-making.