Manipulate to Obfuscate: A Privacy-Focused Intelligent Image Manipulation Tool for End-Users
Kyzyl Monteiro, Yuchen Wu, Sauvik Das · 2024
Image-related privacy preservation techniques often demand significant technical expertise, creating a barrier for end-users. We present a privacy-focused intelligent image manipulation tool that leverages recent advancements in generative AI to lower this barrier. Our functional prototype allows users to express their privacy concerns, identify potential privacy risks in images, and recommends relevant AI-powered obfuscation techniques to mitigate these risks and concerns. We demonstrate the tool’s versatility across multiple different domains, showcasing its potential to empower users in managing their privacy across various contexts. This demonstration presents the concept, user workflow, and implementation details of our prototype, highlighting its potential to bridge the gap between privacy research and practical, user-facing tools for privacy-preserving image sharing.