The present invention provides methods for using abstractions of people, including dynamic and static groups of people, to enhance the efficiency of the specification and automation of policies for sharing information between users with a “need-to-know.” An instance of the present invention can also provide these users information based on a “time-to-know.” By providing access to information based on group affiliation and properties of the content of the information, the present invention maintains optimal information privacy while minimizing encumbrances to sharing data with appropriate users and even at appropriate times. The present invention can be integrated with other communication technologies to facilitate access to information in a time appropriate manner. Other instances of the present invention employ automated and semi-automated, mixed-initiative techniques, to make information-sharing decisions. Additional instances of the present invention include the employment of machine-learning techniques to facilitate construction of access policies from the actions or profile of a single user or a community of users, including the construction of automated sharing agents that work in an automated or mixed-initiative manner to respond to real-time requests for information.