This thesis investigates conflict early warning and early response through the lens of Critical Theory. It argues that the field of conflict early warning has exclusively taken problem-solving approaches by focusing on developing early warning and response instruments such as the indicators for prediction and how to connect warnings with effective responses. On the other hand, it has been unaware of its own theoretical and epistemological foundations. As a result, the studies of conflict early warning have been conducted in a particular manner that works for the interests of liberal western states rather than for those at risk. The thesis argues that the emergence of conflict early warning and response was only possible in the post-Westphalian global order, where the equity of state sovereignty is no longer absolute but conditional, and interventions in internal affairs are justified by appealing for human security. In this global context, unlike the popular belief that conflict early warning is motivated by humanitarianism, conflict early warning emerged as a liberal technique of government that makes those to be governed visible, calculable, comparable and interventionable through data collection and analysis. In other words, conflict early warning systems have been used for the protection and expansion of the western community's security and economic interests. In addition, this particular theory/ideology has marginalized small scale collective violence and local capacity for early warning and response. The thesis regards community-based early warning and response systems as a more emancipatory form of practice in this particular historical moment because they see local actors as the active subject of conflict early warning and early response rather than the mere object of western rescue. Then, this research examines two cases from Kyrgyzstan and Sri Lanka, and identifies their limitations and prospects.