A longstanding open problem in Computer Science is that of how to get high quality statistics through direct queries to databases containing information about individuals without revealing information specific to those individuals. It has long been recognized that the key to making this work is to add noise to query answers. The problem has been how to do this without either adding a great deal of noise to answers or limiting the number of answers an analyst can obtain. This talk presents Diffix, a new framework for anonymous database query. Diffix adds noise in such a way that repeated answers produce the same noise: it cannot be averaged away. This "fixed noise" mechanism, however, creates new opportunities for attacks. Diffix pro-actively tests potential alternate queries to discover and prevent these attacks. In this talk, we describe the Diffix framework and present a system design that provides basic query logic and statistical operations. We will give a brief demo of a more advanced Diffix system that operates as a commercial product.