The Global Reach of Deepfakes

For anyone doubting the global reach or potential harm, both financial and reputational, of generative AI synthetic audio or video, colloquially known as deepfakes, a salient story emerged at this year’s NDTV World Summit in New Delhi.

Global Reach: The Dangers of AI Deepfakes

Addressing the conference, Sunil Bharti Mittal, billionaire, and chairman of Bharti Enterprises, which includes Bharti Airtel, recounted an incident where a senior finance official based in Dubai received a telephone call purportedly from Mr. Mittal requesting a large funds transfer.  The video can be found here.

The official, observing the request was unusual and being diligent, prevented the attempted fraud from being successful. However, Mr Mittal, upon hearing the deepfake audio, admitted he was stunned by the accuracy, including the tone and articulation of the audio.

The Ease in Which a Sopsicaated Deepfake is Created 

So, what would have been required to create this deepfake? Ironically, the video above of Mr. Mittal speaking would provide more than enough audio for anyone else to make another deepfake of Mr. Mittal. The software required is readily available, mostly open source and the quality of the deepfakes is only getting better.

This is also not some isolated incident. A business.com study has found that more than 10% of companies have dealt with either attempted or successful deepfake fraud. Deloitte’s Center for Financial Services predicts that Gen AI could enable fraud losses to reach US$40 billion in the US by 2027. Deepfake fraud detection should be top of mind for all organizations, not just financial services. The US$25 million loss suffered through a deepfake video conference call in Hong Kong actually targeted an engineering firm.

What You Need to Effectively Detect a Deepfake is Available Today

To effectively detect deepfake audio requires AI-based deepfake detection on every corporate audio communication channel including Contact Centres (including IVR/IVA), internal phone networks, enterprise Help Desks, and as well as collaboration tools and online conferencing.

Voice Verity^® is a non-biometric AI solution based on large-scale Deep Neural Network (DNN) techniques. It requires no enrolment, no consent, and no PII. Operating in both real-time and as a background process, it can be integrated into any channel carrying audio and requires just a snippet of audio to analyze and respond with a probability score of that audio being human or synthetic, regardless of language.

Sunil Mittal was fortunate the fraudsters in his case contacted a diligent employee who didn’t automatically follow the instruction he was given, but reliance on human detection or employee diligence is no longer an effective defense against deepfakes. It is now a business imperative to protect every call, every chat, and every conference, and Voice Verity provides the first line of defense in the AI fight against AI.