Worms/Darmstadt, April 15, 2025 – KOBIL, a leading provider of digital identity and mobile security solutions, announces its partnership with the Technical University of Darmstadt to improve the detection of voice deepfakes. Hyper-realistic, AI-generated voices pose a growing security threat – from fraud and identity theft to fake emergency calls that mimic people in distress. As synthetic voices become more common – for example, in chatbots – it is becoming increasingly difficult to distinguish real speech from manipulated content.

To address this challenge, KOBIL is collaborating with the System Security Lab at TU Darmstadt, which has developed an innovative AI-based detection method called VoiceRadar. As part of the partnership, KOBIL will integrate VoiceRadar into its Secure SuperApp platform. The goal is to enhance the trustworthiness of voice communications and provide robust AI-powered cybersecurity.

Ismet Koyun, CEO and founder of the KOBIL Group, emphasizes:
"We live in an age where you can no longer trust everything you see or hear. Deepfake technologies, if misused, can cause immense harm – not just economically, but also socially and emotionally. Traditional security methods are no longer sufficient. VoiceRadar opens up new possibilities to counter emerging threats and make the digital world a safer place."

VoiceRadar sets new standards in voice deepfake detection
The System Security Lab at TU Darmstadt is one of the world's leading research centers in the field of IT security, with a special focus on AI security. The research team has developed a unique deep-learning-based method that analyzes acoustic patterns and frequency anomalies in voice recordings to detect manipulation. By integrating physical modeling approaches – particularly from radar systems – with machine learning, VoiceRadar offers significantly higher accuracy than existing methods when it comes to detecting synthetic voices.

This cutting-edge research was recently honored with the “Distinguished Paper Award“ at the prestigious NDSS-Symposium 2025.

KOBIL Brings VoiceRadar into Real-World Applications
KOBIL is now integrating a tailored version of VoiceRadar into its Secure SuperApp platform, which is already being used in the city apps of Worms and Istanbul, and soon in many other cities. This enables real-time detection of deepfake voices and improves identity verification processes. Users gain an additional layer of protection against identity fraud and social engineering attacks based on AI-generated voice manipulation. Thanks to VoiceRadar, the security and integrity of all interactions via the SuperApp platform are significantly strengthened. Communication, transactions, and authentication processes are protected from manipulation – even by sophisticated voice deepfakes.

"The integration of VoiceRadar into our Secure SuperApp platform marks a major milestone in cybersecurity," says Ismet Koyun.

"Our collaboration with TU Darmstadt is essential to protecting digital identities. I’d like to extend my heartfelt thanks to Professor Ahmad-Reza Sadeghi, Head of System Security, and his highly dedicated and motivated team."

Trust and security in the age of AI
Professor Ahmad-Reza Sadeghi from TU Darmstadt also highlights the importance of the partnership:
"Our research shows that traditional detection methods are reaching their limits. With VoiceRadar, we’re taking a new approach to reliably distinguish AI-generated voices from real ones. The collaboration with KOBIL allows us to apply this technology in practice and create real value for businesses and consumers alike."

The awarded paper can be downloaded here: VoiceRadar – Voice Deepfake Detection using Micro-Frequency and Compositional Analysis