Abstract
In this talk, we overview recent efforts in understanding privacy-utility trade-offs (PUTs) from an information-theoretic perspective. Under certain metrics for privacy and utility (e.g., mutual information and other $f$-divergences), PUTs can be understood specific instantiations of a broader class of formulations called ``bottleneck problems,'' which include the information bottleneck and the privacy funnel. These formulations are closely related to data processing inequalities, and reveal interesting facets of the trade-offs that emerge when designing privacy-assuring mechanisms.