 Believe it or not, about two-thirds of our daily conversations are devoted to social topics, much of which takes the form of gossip. Gossip refers to sharing information about the absent third person between two parties. It is something that we humans do so often, so naturally, that we don't even think of it as anything special. Despite its ubiquity, gossip has received relatively little attention from scientists. In the past two decades, researchers increasingly realize important social functions of gossip. In particular, to the extent that gossip provides information about other's transverseness, it allows the receivers of gossip to detect potential cheaters and selectively cooperate with deserving others. Moreover, when people are unable to directly punish free riders, gossip can be used as an informal, low-cost policing device for controlling free riders and social cheats. Thus, gossip can have enormous impact on people's lives, both by boosting people's reputations and by tearing them down. Accordingly, the threat of gossip by others tends to motivate people to enhance their cooperation in social interactions. However, an important barrier to gossip serving its functions is that people may have the temptation to share dishonest gossip for personal benefit. For instance, in a company where the boss decides to promote employee A or employee B, employee A may share dishonest, negative gossip about employee B if there is a conflict of interest between these two. Examples like this abound in everyday life. But when are people likely to share dishonest gossip? To answer this question, we developed mathematical models to understand when gossip can be honest or dishonest. In these models, we distinguish three different roles, the Gossiper, the Receiver, and the Target. We assume that the Gossiper first shares information about the target's past cooperative or non-cooperative behavior with the Receiver, then the Target has one short interaction with the Receiver. The model also considers the fitness interdependence of the Gossiper with the Receiver and with the Target. These two constructs can be simply interpreted as how much the Gossiper values the Receiver and the Target, or the Gossiper's relationships with the other two. We use four different games to represent various social interactions. Take the Stack Hunt game, for example. Both players have the option to cooperate or defect. It is always beneficial for the ego to behave the same way as the opponent. That is, hunting stack, if and only if one's partner also hunts stack, otherwise, hunt small hair that does not require coordination for success. Thus, only when both decide to hunt the stack can they maximize the collective welfare. Many real-life situations involve coordination like this. Here, sharing honest or dishonest Gossip will influence the Receiver's behavior towards the target in the game, which leads to different payoffs for the Receiver and the Target. Meanwhile, the Gossiper may have some late gains or losses depending on their positive or negative interdependence with the other two parties. When the expected payoff of honest Gossip is larger than that of dishonest Gossip, people will gossip honestly, otherwise, they gossip dishonestly. Through this model, we were able to predict the rational preference for honest or dishonest Gossip strategy in different social interactions. Our model suggests a very simple rule. When there is a perfect match between finished interdependence and the effect of honest Gossip, the Gossiper should always be honest. In case of perfect mismatch between these two, the Gossiper should always be dishonest. For instance, in a stat-hunt game with a cooperating target, honest Gossip benefits both the Receiver and the Target. Therefore, people should be honest when they have positive relationships with both the Target and the Receiver. And dishonest when they have negative relationships with both the Target and the Receiver. What do these results show us? Going back to the previous example about promotion, when an employee gossips to a boss about another employee, if she likes both of them, that it has a positive evaluation and effect of honest Gossip is positive for both of them, then she will be honest, which is shown on the left side. Conversely, if she dislikes both of them but the effect of honest Gossip is positive for both, then she will be dishonest. But what happens if there is an imperfect match between interdependence and the effect of honest Gossip? Well, that is a longer story. You'll have to read our paper for that. Of course, real-life situations are often more complex, but these model results can give us some implications about when people can be honest or dishonest and when others can be trusted. It is unavoidable that people will be sometimes tempted to be honest in their communication about apps and others. The problem of dishonesty and the conditions in which the curse requires further study and some of our author team are conducting additional studies to understand when and why Gossip will be honest or dishonest and how receivers react to different types of Gossip. This is joint work with a large collaboration team without whom this work may not have been born. I would like to thank all co-authors here. Thank you for watching.