I propose a possible development of the Global Neuronal Workspace (GNW) model of access consciousness. Its initial motivation is that the model does not offer a clear distinction between the neural signatures associated with information broadcasting (which is the main function of the GNW) and those of other processes also related to consciousness, such as information integration. I suggest that a theoretically interesting and neurally plausible signature of broadcasting can be provided by using the graph-theoretic approach to information dissemination in communication networks. The theoretical appeal of this framework lies in the fact that, in addition to distinguishing between broadcasting and other relevant communication processes, it can possibly also contribute to identifying the GNW mechanism. I suggest that the approach can provide precise predictions regarding the communication algorithms and wiring diagram that the GNW would implement if it were an efficient broadcasting network.