The problem of statistical disclosure control - revealing accurate statistics about a population while preserving the privacy of individuals - has a venerable history. An extensive literature spans multiple disciplines: statistics, theoretical computer science, security, and databases. This talk revisits the problem of privacy-preserving data analysis from a cryptographic perspective. The focal point of the talk is an _ad omnia (as opposed to ad hoc) notion of privacy, which we call _differential privacy. Roughly speaking, differential privacy ensures that only a limited amount of additional risk _ of anything! -- is incurred by participating in a data set.