Appears in collection : Nexus Trimester - 2016 - Secrecy and Privacy Theme

While quantum cryptography offers interesting security guarantees, it is challenging to build good quantum devices. In practise, we will therefore typically rely on devices built by someone else, whose properties are difficult to characterize and which exhibit inherent imperfections that may threaten security. The goal of device independent quantum cryptography is to develop tests that allow (partially) unknown and uncharacterized quantum devices to be used safely. Specifically, we imagine that quantum devices act as black boxes, which we can only ask to perform certain quantum measurements and collect the resulting measurement outcomes. However, the quantum state or actual measurements performed by the device are unknown. Security guarantees should then be given purely as a function of such classical input/output behaviour. This talk will be an introduction to the concept of device independence in quantum cryptography that does not assume any knowledge of quantum information. We briefly review results in device independent quantum key distribution, before exhibiting very recent work on device independent two-party quantum cryptography. Finally, we will discuss some of the many open challenges in certifying properties of quantum devices based on classically observed statistics.