Online social networks provide an environment for their users to share content with others, where the user who shares a content item is put in charge, generally ignoring others that might be affected by it. However, a content that is shared by one user can very well violate the privacy of other users. To remedy this, ideally, all users who are related to a content should get a say in how the content should be shared. Recent approaches advocate the use of agreement technologies to enable stakeholders of a post to discuss the privacy configurations of a post. This allows related individuals to express concerns so that various privacy violations are avoided up front. Existing techniques try to establish an agreement on a single post. However, most of the time, agreement should be established over multiple posts such that the user can tolerate slight breaches of privacy in return of a right to share posts themselves in future interactions. As a result, users can help each other preserve their privacy, viewing this as their social responsibility. This article develops a reciprocity-based negotiation for reaching privacy agreements among users and introduces a negotiation architecture that combines semantic privacy rules with utility functions. We evaluate our approach over multiagent simulations with software agents that mimic users based on a user study.