Learn how vision-language models can be exploited to infer private attributes from images in this 13-minute research presentation from the University of Central Florida. Explore the privacy implications and vulnerabilities that arise when advanced AI models analyze visual content to extract sensitive personal information that users may not intend to share. Examine the methodologies used to demonstrate these inference capabilities, understand the potential risks to individual privacy in an era of increasingly sophisticated computer vision systems, and discover the broader implications for data protection and AI ethics. Gain insights into how seemingly innocuous images can reveal private attributes through the lens of modern machine learning techniques, and consider the challenges this poses for maintaining privacy in digital environments where visual content is ubiquitous.