We propose a new class of attacks that breach user privacy by exploiting advertising systems offering microtargeting capabilities. We study the advertising system of the largest online social network, Facebook, and the risks that the design of the system poses to the privacy of its users. We propose, describe and provide experimental evidence of several novel approaches to exploiting the advertising system in order to obtain private user information.
The work illustrates how a real-world system designed with an intention to protect privacy but without rigorous privacy guarantees can leak private information, and motivates the need for further research on the design of microtargeted advertising systems with provable privacy guarantees. Furthermore, it shows that user privacy may be breached not only as a result of data publishing using improper anonymization techniques, but also as a result of internal data-mining of that data.
We communicated our findings to Facebook on July 13, 2010, and received a very prompt response. On July 20, 2010, Facebook launched a change to their advertising system that made the kind of attacks we describe much more difficult to implement in practice, even though, as we discuss, they remain possible in principle. We conclude by discussing the broader challenge of designing privacy-preserving microtargeted advertising systems.
"Privacy Violations Using Microtargeted Ads: A Case Study,"
Journal of Privacy and Confidentiality: Vol. 3
, Article 3.
Available at: http://repository.cmu.edu/jpc/vol3/iss1/3