To investigate the effects of private communication on support for prosocial collective choices, we conduct a laboratory experiment in a public goods setting with a majority vote, manipulating whether participants can have private conversations before a public discussion. When private conversations are allowed, majority tyranny outcomes are more common and prosocial collective choices less common. In both communication treatments, participants start out proposing and voting for prosocial allocations and increase their support for majority tyranny allocations as the experiment progresses. Private communication therefore increases the prevalence of majority tyranny (but is not strictly necessary for it). The results demonstrate that the effects of private communication in multilateral bargaining extend to public goods settings, despite the contrasting finding in the latter that democratic institutions enhance prosocial outcomes.