For blind and low-vision individuals, visual assistant applications such as BeMyAI and SeeingAI provide help with daily tasks by answering questions about images. However, these tools often require users to capture photos that may include sensitive information, like account numbers or personal documents. While these applications warn against sharing personally identifiable data, it can be difficult for users to avoid exposing private details.
Researchers from Stony Brook University, the University of Texas at Austin, and the University of Maryland have developed a new framework called FiG-Priv to address this issue. The system is designed to protect user privacy by selectively hiding only high-risk information in images submitted to AI assistants. Details such as Social Security digits or credit card numbers are concealed, but less sensitive context—like the type of document or a customer service number—remains visible.
Paola Cascante-Bonilla, assistant professor in the Department of Computer Science at Stony Brook University and co-author of the study, described the approach: “Traditional masking techniques to protect sensitive information often blur or black-out entire objects. For blind and low-vision users, this is impractical. Masking too much destroys the utility of the content, while masking too little leaks sensitive data. FiG-Priv aims to allow BLV users to interact with AI systems without exposing personal information. It focuses only on the sensitive content.”
FiG-Priv works by detecting and segmenting private objects within an image—such as financial statements or credit cards—and then redacting those areas with black squares. The rest of the image remains clear so that visual assistants can still interpret useful details.
Lead author Jeffri Murrugarra-Llerena emphasized how this solution supports both privacy and independence: “Blind and low-vision users should be able to support both their independence and their privacy. In previous approaches, they were forced to choose one over the other. With our approach, users can ask questions more confidently, without worrying about what these systems might reveal.”
The full story by Ankita Nagpal is available at the AI Innovation Institute website.