SN2K Attacks and Honest Services

In this paper, we define and illustrate a new form of attack in the context of software services: the software-based need-to-know (SN2K) attack. SN2K attacks can be carried out by dishonest provider of a software service so that it can maliciously gain access to sensitive information, even if the service does {\em not need to know} such data in order to compute the functionalities offered by it. We prove that it is generally undecidable to detect whether a given implementation of a service is dishonest, i.e., it implements an SN2K attack. A certification scheme for honest services is also proposed; our scheme relies on program slicing and certain other aspects of static program analysis.