Perceiving emotions in others is at the foundation of higher-order social cognition. The importance of emotions is evidenced by the fact that they receive prioritized attention at early stages of processing the environment in humans and some other primates. Nevertheless, we do not fully understand how emotion modulates attention over longer durations in primates, particularly in great apes. Bonobos, one of our closest relatives, stand out in emotion processing and regulation among great apes. This makes them an interesting comparison species and a valuable model for studying the evolution of emotion perception in hominids. We investigated how bonobos and humans spontaneously attend to emotionally valent scenes in a preferential looking task using eye-tracking. With Bayesian mixed modelling, we found that bonobos and humans generally looked longer at emotional scenes, mainly of conspecifics. Moreover, while bonobos did not have a bias toward emotional human scenes, humans sustained their attention toward bonobos playing, grooming and having sex. Furthermore, when exploring an immediate bias for emotions, humans showed a bias toward affiliative human scenes, and bonobos showed a bias away from bonobos-in-distress scenes. These findings suggest that emotions modulate attention at early and later attentional stages in bonobos, similar to humans.