Red Bleed - A Pragmatic Near-Infrared Presentation Attack on Facial Biometric Authentication Systems

Explore a groundbreaking cybersecurity research presentation that exposes critical vulnerabilities in near-infrared facial biometric authentication systems through a novel attack method called "Red Bleed." Learn how researchers from Nanyang Technological University demonstrate how widely-used commercial facial recognition systems like Windows Hello and Apple FaceID can be compromised using a custom-built LCD display costing less than $400. Discover the technical methodology behind converting visible spectrum RGB images into near-infrared domain attacks using a novel Variational Autoencoder framework combined with advanced face swapping techniques, enabling attackers to bypass NIR-based authentication using readily available social media photos or video conference captures instead of requiring live NIR video samples. Understand the practical implications of this vulnerability that affects commercial-off-the-shelf enterprise-grade face authentication systems, examine the hardware design and implementation details, and review the responsible disclosure process including Microsoft Security Response Center confirmation and scheduled CVE release for June 2025.