Smartphone features such as tilt control, touch screens, and dynamic screen brightness rely on sensor data that many users consider benign, but research has demonstrated that this data can be exploited for privacy invasions through inference attacks. These attacks analyze sensor data to extrapolate sensitive information, leveraging inertial measurement units (IMUs) to infer activities or interactions. Studies have shown how data from gyroscopes, accelerometers, and magnetometers can be used to deduce keystrokes on touch screens and even eavesdrop on conversations by exploiting the reverberations from smartphone loudspeakers. Inference attacks have also been demonstrated on smartwatches, with the ability to infer typed words based on wrist movements. As devices become more sensor-rich and machine learning tools advance, these attacks are likely to increase in accuracy and scale, urging reconsideration of sensor data's role in privacy and security.