Quantum information theory studies fundamental limits on the ways that physical systems can be manipulated. In one of the most general formulations, systems are represented by C*-algebras, and transformations by completely positive maps. The Stinespring dilation theorem - which fully characterizes these maps - provides some of the most basic results in quantum informations. I will explain its role in quantum error correction and the information-disturbance tradeoff. Other results in quantum information rely on an additional bit of physics: a notion of locality. It is needed, for instance, in order to define the concept of entanglement. I will give example of "exotic" locality structures, such as that of fermions.