Within a mobile device, a method and system to produce a probability the mobile device is in possession of a known person, the first user. Sensors are used to detect and quantify the behavioral biometrics of the human traits of the person in possession of the device. On a continuous basis, a machine learning process collects the biometrics of several traits of the first user memorizing the artifacts of neural networks used for learning. Subsequently, a prediction neural network provisioned with these artifacts and processing new biometric inputs of the present user of the device produces a probability the present user of the device is the first user. Affirmation of identity can then be made based on that probability.