In some instances, a document may be worn or poorly illuminated, while the user may hold it at an angle or move the camera during capture. Those conditions often prevent software from reliably locating and reading the machine readable zone (MRZ). The commercial challenge is achieving accurate first-pass extraction across different document types, devices and lighting conditions. Recognition occurs locally within mobile apps and web browsers, allowing document data to be extracted without sending images to the cloud. However, on-device processing alone does not determine privacy compliance. Organizations deploying the software may still transmit, store or process document images and extracted data as part of their identity verification workflows