"US Census Data Vulnerable to Attack Without Enhanced Privacy Measures"
Computer scientists at the University of Pennsylvania School of Engineering and Applied Science designed a "reconstruction attack" that demonstrates the vulnerability of US Census data to exposure and theft. Aaron Roth, Henry Salvatori Professor of Computer and Cognitive Science in Computer and Information Science (CIS), and Michael Kearns, National Center Professor of Management and Technology in CIS, led a recent PNAS study demonstrating that statistics released by the US Census Bureau can be reverse-engineered to reveal confidential information about individual respondents. The study team identified risks to the privacy of the US population using computer power no stronger than that of a commercial laptop and an algorithm design based on Machine Learning (ML) concepts. The study is notable for being the first of its kind to establish a baseline for unacceptable vulnerability to exposure. In addition, it demonstrates that an attack can help determine the likelihood that a rebuilt record corresponds to the data of a real person, making it more likely that this type of attack could expose respondents to identity theft or discrimination. This article continues to discuss the reconstruction attack demonstrated to prove the vulnerability of US Census data to exposure and theft.