Medical research relies heavily on Big Data - large volume of training and testing data is required to build an automatic medical diagnostic tool using advanced Machine Learning techniques. However, presence of Protected Health Information (PHI) in medical images constitutes a significant hurdle to sharing and aggregating data. Due to sheer volume of the datasets, manual removal of PHI is not scalable and building a tool for automatic PHI detection, deletion and masking is crucial. If PHI is leaked there are financial and legal ramifications (HIPAA, GDPR), but the biggest concern is reputation damage. To address this problem, Glendor has developed PHI Sanitizer, an AI-based system for fully automatic customer premises-based software that sanitizes PHI information from medical images and metadata.