Cedric Langbort University of Illinois at Urbana-Champaign The emerging paradigms of urban sensing and analytics in "smart cities" call for a refined understanding of how distributed sensing, communication, information transmission, and machine learning perform in the presence of strategic self-interested sensors and data sets, and the development of new tools, algorithms, and protocols that can withstand them. We believe that the traditional tools of cheap talk theory (which considers scenarios of strategic interaction between an informed sender and a decision-making receiver) can help shed new light on many of these issues and, more generally, on security problems in cyber-physical systems. After reviewing the classical framework and results of cheap talk theory in Economics, we will present new models of strategic information transmission inspired by them, with some applications to CPS security.