When: Tuesday, 17 March 2026 | 17:00 – 18:30 CET
Where: Hasso Plattner Institute, Campus Griebnitzsee, Building L (Campus Kitchen One), Room L-1.02, 14482 Potsdam
Host: AI Service Center Berlin-Brandenburg @ HPI
About the Event: Deepfakes have been a subject of heated debate for some time. However, with the advent of newer and more powerful generative AI models, distinguishing authentic content from artificially generated images and videos is becoming increasingly difficult. This raises critical questions: What distinguishes “fake” images from “real” ones? How are Deepfakes generated to look so deceptively authentic? And most importantly: How can we technically identify them when the naked eye is no longer sufficient?
In this guest talk, Jonas Bernhard will provide an in-depth insight into the topic, introducing various methods for both the generation and detection of manipulated media. As a concrete approach, he will present Clyravision—a multi-agent system that combines different models and techniques to detect Deepfakes in an interpretable manner.
About the Speaker: Jonas Bernhard is a Machine Learning Engineer at Merantix Momentum. He studied at the Karlsruhe Institute of Technology (KIT) and TU Berlin, specialising in Information Systems, Machine Learning, and Computer Vision. In his work, he and his colleagues are committed to the responsible use of AI to solve complex visual challenges across a wide range of projects.
Registration: Participation is free of charge, but binding registration is required.
