Super-sticky and declassifiable release policies for flexible information dissemination control

Abstract

Over the years, many aspects of the transfer of information from one party to another have commanded the attention of the security and privacy community. Released information can have various levels of sensitivity: facts that are pub-lic, sensitive private information that requires its original owner's permission for its future dissemination, or even in-formation that requires control over the release of the con-clusions reached using that information. Some situations also call for declassification of information, which requires a two-pronged approach: the original owner retains control over the dissemination of sensitive information and sensitive conclusions reached using that information, but when the in-formation is used to reach conclusions that are sufficiently non-sensitive, the original owner's control can be removed for the dissemination of those conclusions. In this paper, we define such a logic to specify information dissemination con-trol policies and reason about release and declassification, and give case studies of the use of our language to control the release of aggregated open source software, multimedia content and medical information.