The term affective computing was coined twenty years ago to refer to computers human-like capabilities to detect and recognize user’s emotions.  Mobile sensing systems can be used to sense the emotional state of one or more users and let a third-party can use this information to produce changes in the user’s emotional state, or analyze hundreds of thousands of pictures, gestures, speechs and so on of people and train recognition systems for affective computing applications. For example teachers or e-learning systems as third-party systems can react appropriately maintaining motivation for their students according their emotional states using augmented reality techniques or changing the multimedia resources used in the lectures. Despite the direct benefits of knowing the emotional states, people in general is opposed to a system captures their emotions with smart phones equipped with built-in or external sensors such as image sensor to capture images and record videos, or a pressure sensor  to detect the force or rhythm of  a finger or stylus pen strokes  within the display area. For that reason, currently privacy is one of the important barrier that limits the social acceptance of mobile sensing systems to do affective computing. In this paper we focus on mobile sensing systems to do affective computing preserving the user’s privacy to motivate the users to be sensed.