Daon launches xSentinel to fight voice deepfakes

xSentinel is an expansion of the company’s proprietary technology, AI.X, and an agnostic security tool aiming to analyse incoming voice channel communication to determine whether the audio comes from a real human or is synthetically generated. Daon’s tech solution targets identity fraud and wants to combat deepfakes, an uprising phenomenon, especially among fraudsters.

The dangers of deepfakes

Deepfake technology has rapidly advanced in the past couple of years, and now can successfully mimic voice signals that previously provided identification information about a caller’s characteristics, including tone and articulation. Using the right GenAI tools, scammers and fraudsters can now generate synthetic voices that mimic human ones, perpetuating the cycle of scams, especially for bank institutions who use voice recognition as an extra layer of biometric security. 

According to internal research by Daon, 62% of IT executives in B2B businesses are concerned about the security threats posed by AI and deepfakes, with several Tier-one executives from global banks having their identity credentials stolen or cloned in the past year due to flawed authentication systems implemented internally.

How does xSentinel work?

The new fraud solution deployed by Daon generates a signal to potential fraud that ranks higher in accuracy than traditional audible cues, effectively eliminating the advantages of cloned voice generators used by bad actors. Once a call is initiated, the propriety algorithms detect various cues within seconds and efficiently distinguish between digitally generated voices and human voices. xSentinel provides the necessary data for companies to build reliable synthetic voice protection steps and add extra layers of security to protect their clients’ data and privacy when transacting online. 

xSentinel leverages AI-powered synthetic voice detection tools built into its system to keep pace with the latest developments of the AI technology and the ever-challenging fraud landscape. By deploying strong models of machine learning that continuously adjust and improve detection patterns, xSentinel proves an efficient tool in combating fraud attempts in real time.

According to Daon officials, the new fraud detection tool represents part of the company’s integrated efforts to manage the balance between a good, frictionless user experience and security. It detects attacks coming from the misuse of GenAI (generative artificial intelligence), as part of the AI.X suite of tools to detect and combat deepfakes.

xSentinel works as a real-time, affordable solution that helps identify potential fraud with a series of key features, including the fact that it is universally applicable – being both language and dialect agnostic, its seamless implementation, and real-time signalling for early, in-call detection of deepfakes and synthetic identities.

About Daon