Author:

Keywords:

iMinds

Abstract:

File sharing services are used daily by tens of thousands of people as a way of sharing files. Almost all such services, use a security-through-obscurity method of hiding the files of one user from others. For each uploaded file, the user is given a secret URL which supposedly cannot be guessed. The user can then share his uploaded file by sharing this URL with other users of his choice. Unfortunately though, a number of file sharing services are incorrectly implemented allowing an attacker to guess valid URLs of millions of files and thus allowing him to enumerate their file database and access all of the uploaded files. In this paper, we study some of these services and we record their incorrect implementations. We design automatic enumerators for two such services and a privacy-classifying module which characterizes an uploaded file as private or public. Using this technique we gain access to thousands of private files ranging from private and company documents to personal photographs. We present a taxonomy of the private files found and ways that the users and services can protect themselves against such attacks.