Aug
27
Date: 27-August-2026
Time: 8:00 AM ET (New York Time)
Presenter: Dr. Siwei Lyu
Abstract
Rapid advances in generative AI have made it increasingly easy to synthesize or manipulate faces, voices, and real-time audiovisual streams, creating new challenges for biometric identity systems. This talk examines how deepfakes can affect different stages of the biometric pipeline, including synthetic identity creation, impersonation, morphing, and real-time digital injection. It will clarify the complementary roles of biometric matching, presentation attack detection and liveness, and media forensics, and review the forensic evidence that current deepfake detectors exploit.
The talk will also discuss why deepfake detection remains difficult in real-world settings and argue for moving beyond a single “real or fake” detector toward layered, evidence-based and risk-aware biometric authentication. The webinar should be relevant to researchers, practitioners, students, and security professionals working in biometrics, identity verification, media forensics, AI security, fraud prevention, and trustworthy AI.
Biography
Dr. Siwei Lyu
Siwei Lyu is a SUNY Distinguished Professor at the University at Buffalo, State University of New York, where he directs the Institute for AI and Data Science and the UB Media Forensics Lab. His research focuses on digital media forensics, deepfake and AI-generated media detection, and trustworthy AI. He leads the development of DeepFake-o-Meter, an open platform integrating state-of-the-art deepfake detection methods for research and real-world analysis. He is a Fellow of IEEE and IAPR.
