Power can be obtained through brute force but also by different social strategies that ensure alliances. This maneuvering in a complex social world is hypothesized to be paramount in the evolution of intelligence. Most empirical evidence for this hypothesis stems from primate research and mostly focuses on the formation and maintenance of coalitionary social bonds. However, little is known about how coalitionary support may be manipulated, and how third-party individuals may hinder the formation of such coalitionary social bonds. Here, we investigated interventions in affiliative interactions of others in a group of 94 individually marked wild ravens, Corvus corax. We show that these interventions occur regularly and are not without risk. Moreover, we show that the identities of both intervener and intervened pairs are not randomly distributed and do not follow rules based on risk assessment. Instead, we observed that ravens with already existing alliances initiate most interventions, and that ravens that are creating new alliances are more likely to be the target of such interventions. These data suggest that high-ranking individuals are using interventions to prevent others from forming alliances and consequently rising in rank. Although being described for the first time, we feel that that the examination of these intriguing social maneuvers has great potential in other social species as well. Finally, in our talk we will discuss the possible cognitive mechanisms behind this behavior.