Over the last decade Online Social Networks (OSN) privacy has been thoroughly studied in many aspects. Some of these privacy related aspects are trust and credibility involving the OSN user-data conveyed by different relationships in the network. One of OSN major problems is that users expose their information in a manner thought to be relatively private, or even partially public, to unknown and possibly unwanted entities, such as adversaries, social bots, fake users, spammers or data-harvesters. Preventing this information leakage is the target of many OSN privacy models, such as Access Control, Relationship based models, Trust based models and many others. In this paper we suggest a new Role and Trust based Access Control model, denoted here as RTBAC, in which roles, that manifest different permissions, are assigned to the users connected to the Ego-node (the user sharing the information), and in addition, every user is evaluated trust wise by several criteria, such as total number of friends, age of user account, and friendship duration. An interesting extension of the model of image anonymization is also given, where a user that has a certain role with a proper permission can access a partial instance of the data, if a sufficient trust level is not achieved. These role and trust assessments provide more precise and viable information sharing decisions and enable better privacy control in the social network.