The objective of MuDDi is to develop robust, multi-modal approaches for detecting media manipulation, combining computer vision, signal processing, and artificial intelligence. The system analyzes micro-movements in skin through video plethysmography (VPG) to detect absent or unnatural pulse waves in manipulated videos. Simultaneously, it employs advanced image forensic methods and invisible watermark analysis to identify tampering in high-resolution media. In addition, artificial expert systems, based on large language models, provide AI-powered assessments of media authenticity, complementing human expert judgment.

MuDDi is designed for professional applications in digital forensics, fact-checking, and law enforcement. It supports media organizations and fact-checkers in verifying video and image content, assists security authorities in the collection and analysis of digital evidence, and helps technology companies and platforms combat disinformation. The system has already been applied in real-world scenarios, such as the CORRECTIV Faktenforum, a participatory platform that allows citizens and experts to collaboratively identify manipulated content and misinformation.