"Anonymous Data Doesn't Mean Private, Researchers Say"

A team of researchers at the Illinois Institute of Technology used Machine Learning (ML) and Artificial Intelligence (AI) algorithms to extract personal information, specifically protected characteristics such as age and gender, from anonymous cell phone data, thus raising concerns about data security. The researchers successfully estimated the gender and age of individual users through their private communications using data from a Latin American cell phone company. They created a neural network model that estimates gender with 67 percent accuracy, significantly outperforming modern techniques such as decision tree, random forest, and gradient boosting models. Using the same model, they were also able to estimate the age of individual users with a 78 percent accuracy rate. Although age and gender information appear to be innocuous, people can use it in malicious ways, often with disastrous consequences. When someone with bad intentions targets young children for anything from sales to sexual predation, they are breaking a number of laws designed to protect minors, including the Children's Online Privacy Protection Rule (COPPA) and the Health Insurance Portability and Accountability Act (HIPAA). Seniors can be targeted by sophisticated spam and phishing efforts due to their susceptibility and access to savings. This data was extrapolated using readily available computing equipment. The neural network model was run on a Linux (Fedora) operating system with 16 GB of memory and an Intel i5-6200U CPU with four cores. The laptop they used for this work is far from exclusive. A well-resourced adversary will have access to much more powerful machines, including cluster computing, in which multiple computers are configured in a cluster to provide computer power for the AI/ML models. This article continues to discuss the team's demonstrated extraction of personal information from anonymous cell phone data using ML and AI algorithms. 

IIT reports "Anonymous Data Doesn't Mean Private, Researchers Say"