Deep learning changes one’s perspective on how the world works and, by extension, how one should be in the world. Thus, whenever new knowledge threatens to disturb the existing political equilibrium, learning becomes a political act. This chapter considers the work of Paulo Freire, Myles Horton, Michel Foucault, and John Dewey, and ends with four recommendations for making political dynamics more conducive to deep learning: identify existing systems of power relationships, surface hegemonic assumptions, speak truth to power, and cultivate procedural justice.