The existing security paradigm does not work for mobile and wearable computing due to limitations in battery power, computational resources and user interfaces; and is also commonly impacted by constraints imposed by carriers. Using Google Glass as an example platform, how can a user input a credential without exposing it to observers? How can the system protect a user against information overload and spam — a problem that is only going to increase in importance as the user is constantly exposed to messaging? And how can we protect users against malware and service providers against digital right breaches, such as jailbreaks and unpaid tethering?

The CRASH/SAFE project is building a network host that is highly resiliant to cyber-attack. One pillar of the design is pervasive mechanisms for tracking information flow. At the lowest level, the SAFE hardware offers fine-grained tagging and efficient support for propagating and combining tags on each instruction dispatch. The operating system virtualizes these generic facilities to provide the information-flow abstract machine on which user programs run. In this talk, we'll take a guided tour of (a simplified model of) the SAFE hardware and software and an end-to-end proof of noninterference for this model.