Human-wildlife conflict is a widespread problem that threatens both animals and local livelihoods. When this conflict involves predators, the additional risks to human life worsen the problem. The perceptions of human-predator conflict have been investigated multiple times, usually to generate data for conflict resolution. However, such efforts have largely involved mammals from the order Carnivora. In this work, we tested hypotheses about the feeding habits of two species of constricting anacondas from the genus Eunectes and human perception of risks and attitudes associated with such habits. In particular, we investigated the consequences for the snake after predation on domestic or wild animals. We also tested the relationship between human killing of anacondas and local development indicators – a proxy for general education level. Data were obtained from internet videos using a standardized method for information retrieval. A total of 330 videos of human-anaconda encounters were recovered from ten different South American countries. We found that visual evidence of a recent meal (distended abdomen) and predation on domestic animals did not affect the probability of the anaconda being killed, but this probability increased as the Human Development Index (HDI) diminished and as anaconda size increased. Although retaliatory killing is described as one of main causes of animal mortality following human-wildlife conflicts, our results suggest that killing of anacondas are not retaliatory or related to economic losses, but preventive, because these snakes are seen as life-threatening animals. Adding social context information on human-predator conflicts, together with taxon-free research on this subject can improve our knowledge about human-wildlife conflict. Further research could focus on the actual impacts (direct or indirect) of anacondas on stakeholders, and on the role of other non-mammalian predators in human-wildlife conflicts (whether actual or perceived).