The view over the city in the January light was definitely real – but you couldn’t be so sure about the pictures on the screen. On January 13 and 14, 2025, the twelve teams of the first stage of the SPRIND Funke Deepfake Detection and Prevention came together for their first Challenge Days in Leipzig. The aim of the Challenge Days is to network and exchange ideas with each other as well as with experts and relevant stakeholders from politics, business, and society.

The event naturally kicked off with the most important players in the competition itself: The teams presented their respective approaches to the other participants, thus directly opening the door to a lively exchange and possible cooperation. All teams in the first stage are working on marking and/or recognizing deepfakes, but are otherwise characterized by great heterogeneity in terms of both their background and their approach. The potential customers that the participants want to reach with their applications are just as diverse.

But for which target group is the technology relevant? What requirements do potential customers have and how do they differ? Harmen Zell (META), Louisa Well (Bündnis 90/Die Grünen), and Stefan Voss (dpa) answered these and other questions in a panel discussion that shed light on the customer perspective and provided the teams with important ideas for their product-market fit. Arne Petersen from SPRIND commented on the presentation from the perspective of an experienced UX designer.

Serial entrepreneur and SPRIND innovation manager Mirko Holzer offered insights into the topic of How do I found and run a successful company? to the teams and new founders, some of whom are still academics. Moritz Griesel from the University of Göttingen rounded off the event with a look at the complexity of regulation on deepfakes and AI in general within the European Union and beyond.

There were further opportunities for the teams to discuss the core assumptions of their business model with each other and with the experts in small workshop formats and over dinner. The participants can now use these findings to further advance their technology in the run-up to the jury meeting for the second stage of the Funke in May 2025.

Following a successful application process, the teams started the Funke in November 2024. The Funke has a total duration of 13 months. The first stage will last seven months and have a maximum financing amount of EUR 350,000 per team; the second lasts six months and is financed with up to EUR 375,000.

ABOUT THE FUNKE

The aim of the Funke is to develop a comprehensive prototype that includes reactive measures for detection and/or preventive measures to protect against image deepfakes and that can be integrated into existing digital infrastructures. The prototype must demonstrate how image deepfakes can be reliably recognized and authenticated. The entire system can work with AI support and must be able to continuously adapt to new deepfake techniques. At the end of the process, at least three different use cases should be covered, for example social media, news portals, and video conferencing systems. Scalability and adaptability to different digital platforms should be possible.

The topic is of great importance: Manipulated, falsified, or out-of-context images undermine trust in visual information, make it difficult to verify content, and can have a significant negative impact in journalism as well as in areas such as politics and business. Technological approaches are therefore crucial to ensure the authenticity of images and videos.

The focus here is not only on minimizing risks and combating dangers, but also on making sensible use of the opportunities that AI technologies open up.