German anti-deepfake effort takes on another fighter

Germany’s federal government is expanding resources for a multi-year deepfake detection project that it is funding.

Executives with BioID, a German biometric anti-spoofing vendor, say the firm has joined a consortium of organizations seeking effective methods of unmasking fraudulent, AI-based images, video and audio.

The company is joining several research organizations in the consortium, including the Fraunhofer Institute for Telecommunications’ Heinrich Hertz Institute, digital ID firm Bundesdruckerei and the Berlin Institute for Safety and Security Research. The project is funded by Germany’s Federal Ministry of Education and Research (BMBF).

According to BioID, the consortium has an April 2024 deadline to come up with something, likely software.

BioID is contributing anti-spoofing technologies and biometrics expertise to the effort, dubbed ‘FAKE-ID.’ Captured motion from two facial images, analyzed with artificial intelligence, enables BioID’s liveness detection to identity deepfake presentation attacks, the company says.

CTO Dr. Robert Frischholz says the machine learning methods used to create deepfakes are similar to some parts of its biometric liveness detection solution.

The consortium might end up going significantly upriver in AI development looking for solutions. The future of detecting deepfakes is being hampered by biased datasets, according to University of Southern California researchers.

Researchers with Wichita State University released a paper this year saying that biometric facial recognition could help in the effort. The technology removes the need for huge quantities of fake data for model training. It also could help bring generalizability to the fight against ever-evolving methods for making deepfakes.

Article Topics

AI  |  BioID  |  biometric liveness detection  |  biometrics  |  biometrics research  |  Bundesdruckerei GmbH  |  deepfake detection  |  deepfakes  |  Germany  |  machine learning  |  presentation attack detection  |  spoof detection