Synthetic versus Real Media: From the age of signal processing to the battle between AI models

Over the last few decades, the realism of synthetic media has increased dramatically. The multimedia research community has developed techniques to distinguish real and fake media, initially focusing on images and, more recently, videos. Early methods relied heavily on signal processing and artefact detection. However, in recent years, AI-generated media has become so hyper-realistic that it is perceived as “more real than real.” This has sparked the AI-versus-AI battle towards new defensive models.

Bio

Giulia Boato is a Full Professor at the Department of Information Engineering and Computer Science, University of Trento, Italy. Since 2012, she has pioneered research at the intersection of signal processing, physiological signal analysis, and, more recently, advanced deep learning methods to distinguish between virtual and real humans. Her work also addresses various forms of digital media manipulation, with a recent focus on deepfake detection and forensic analysis in open-world scenarios such as social media. She has authored over 140 publications in international journals. Her research spans image and signal processing, multimedia data protection, and digital forensics. She is an elected member of both the IEEE Multimedia Signal Processing Technical Committee (MMSP TC) and the IEEE Information Forensics and Security Technical Committee (IFS TC).