High levels of hostility between individuals on opposing sides of politics have led to a burgeoning literature on the concept of affective polarisation. Though also explained by ideological disagreements, intra-citizen hostility is primarily explained through the concept of social identities, and, in particular, partisanship. However, the number of partisan identifiers have either remained stable or declined in advanced democracies, and levels of party identification are no stronger than before. Why, then, is politics so divisive? I tackle this puzzle by focusing on the rise of negative partisanship, a sense of 'anti-them' identity that has come to structure political behaviour in advanced and new democracies like. I leverage survey data collected on the eve of the last German elections (N = 2000) to provide a robust measure of negative partisanship as a social identity, and show its effects on out-voter dislike via an experiment that independently triggers positive and negative party identities via news media content. I compliment the experiment with web-tracking data and automated text analysis to analyse respondents' media diets using an innovative word embedding method, allowing for article-level content analysis. I expect negative partisans to display higher levels of out-voter dislike and to disproportionately consume media content that attacks the party with which they negatively identify. This interdisciplinary paper is the first to combine the concept of negative partisanship and online news media consumption, empirically and theoretically advancing our understanding of negative political identities and their effects on affective polarisation.