Biometric specialist Aurigin has successfully integrated its advanced voice liveness detection technology into Swisscom’s customer identification infrastructure. The deployment is a direct response to the operational imperative of countering AI-generated audio deepfakes, which have rendered traditional voice recognition insecure.
As Swisscom relies increasingly on remote identification for its Sign service and call center operations, the threat of injection attacks, where fraudsters feed synthetic audio directly into a data stream, has grown. Aurigin’s solution works by analyzing the micro-spectral features of a voice sample in real-time. Within approximately 15 seconds, the system can distinguish between the complex acoustic signature of a live human vocal tract and the flatter signature of a synthetic clone or recording.
This technology allows Swisscom to maintain compliance with strict mandates from FINMA (the Swiss Financial Market Supervisory Authority) regarding the proof of liveness for digital account opening. “Fraudsters no longer need stolen documents; an online video or short audio clip is often enough,” a Swisscom spokesperson noted, highlighting the necessity of this AI-driven defense layer.
The integration outfits Swisscom’s Sign identification process with audio liveness verification and voice fingerprinting at the moment of identity validation. The company says extending verification beyond onboarding turns a regulatory requirement into a continuous, low-friction trust framework, one that reinforces security with every interaction and aligns with EU and FINMA mandates for auditable liveness verification.
As deepfake attacks continue to surge globally, with voice-based fraud becoming increasingly sophisticated, Aurigin’s Apollo model achieves over 98 percent accuracy with latency of less than 50 milliseconds, making it suitable for real-time fraud prevention in high-stakes identity verification scenarios.
—
By the ID Tech Editorial Team
