News

11/12/2024

Fight Against Digital Deception

Jury selects FZI project for deepfake detection in SPRIND innovation challenge

In the first step, 12 teams will receive up to 350,000 euros for their work on developing solutions to effectively detect deepfake photos, amongst them the FZI team with its submission “MuDDi: Multi-modal Deepfake Detection”.

In the 21st century, the integrity of digital media and information is increasingly threatened by so-called deepfakes – highly realistic, AI-generated synthetic content. This technology carries significant risks for disinformation, fraud, and manipulation of public opinion. To address this challenge, the Federal Agency for Disruptive Innovation SPRIND is running the SPRIND Funke “Deepfake Detection and Prevention” innovation challenge on behalf of the Federal Ministry for Digital and Transport (BMDV).

Multi-modal Deepfake Detection

One of 12 award-winning teams in the innovation competition is the FZI team MuDDi, short for “Multi-modal Deepfake Detection”. The approach of the FZI team around Dr. Jonas Fegert, Dr. Steffen Thoma, Prof. Dr. Achim Rettinger, and Prof. Dr. Wilhelm Stork: To counteract the influx of synthetic data, the MuDDi project is developing a comprehensive approach for detecting deepfakes in different media types, in other words, for videos as well as for images and audio.
The aim is a comprehensive probability assessment for the detection of deepfakes across multiple media types.
In image analysis, for example, the project focuses on high-level features such as geometric patterns and color distributions, which are supplemented by methods for detecting watermarks.

Project manager Dr. Jonas Fegert: “This is part of our contribution to a more resilient democracy and a novel approach from our research at the FZI on Applied Artificial Intelligence and Digital Democracy and Participation.”

New insights and methods in AI and machine learning for better deepfake detection

Scientific progress in recent years has led to new findings and methods in Artificial Intelligence and machine learning. These can be used to develop more robust deepfake detection and prevention systems. Although various approaches to combating deepfakes have already been demonstrated, these have often been limited to niche applications or specific types of deepfakes. However, the aim must be to develop comprehensive and scalable solutions to ensure trust in our media and communication.

There are several challenges to overcome regarding adequate protection against deepfakes. For example, deepfake detection is often unreliable as synthetic content creation technologies are rapidly evolving. Existing detection methods have difficulties with generalizability and are often only effective for certain deepfakes. Prevention measures are limited and cannot seamlessly integrate into existing digital infrastructures.

“The SPRIND Funke Deepfake Detection and Prevention aims to develop solutions for the effective detection of deepfake photos and prevention through the non-manipulability of original photos. The solutions to be developed should take into account the entire value chain of digital media processing – from preventive authentication and securing of original content, real-time analysis and detection of manipulated or synthetically created media, to seamless integration into existing digital infrastructures and platforms,” explains Dr. Jano Costard, Challenge Officer at SPRIND.

About the SPRIND innovation challenge

An expert jury has now selected a total of 12 teams from almost 50 applications to receive up to 350,000 euros each for the next seven months. The teams come from research institutions, universities and start-ups from France, Germany and Slovenia.

The teams will present to the jury again in May 2025. The jury will then select up to 10 teams, which will receive up to an additional 375,000 euros per team for the further development of their solution until the end of the innovation competition in November 2025.
For more information and descriptions of the participating teams, please visit
https://www.sprind.org/en/impulses/challenges

About the FZI

The FZI Research Center for Information Technology, with headquarters in Karlsruhe and a branch office in Berlin, is a non-profit institution for information technology application research and technology transfer. It delivers the latest scientific findings in information technology to companies and public institutions and qualifies individuals for academic and business careers or the leap into self-employment. Supervised by professors from various faculties, the research groups at the FZI develop interdisciplinary concepts, software, hardware and system solutions for their clients and implement the solutions found as prototypes. The FZI House of Living Labs provides a unique research environment for application research. The FZI is an innovation partner of the Karlsruhe Institute of Technology (KIT) and strategic partner of the German Informatics Society (GI).

Press contact

Valérie Hasler

Communications

Further links: