The Rise of E-commerce Fraud Through AI

A Juniper Research study, “Global Merchant Fraud Prevention Market 2024 – 2029”, concludes that money lost to online (e-commerce) fraud will increase by 141% between now and 2029, rising from $44bn to $107bn over this period.

E-commerce Fraud Soars: AI Deepfakes and the Battle for Merchant Security

A major reason they attribute to this increase is the role of AI particularly deepfakes, to defeat verification systems used by e-commerce merchants. Their recommendation is for merchants to integrate with fraud prevention systems that contain AI to detect deepfakes and other emerging tactics from fraudsters.

The use of generative AI deepfakes in attempting to bypass verification systems, whilst increasing rapidly, is not new. Numerous attempts have been reported worldwide, some successful, and some not, against both biometric-based identity verification solutions as well as human-based verification, such as recognizing a colleague over an online meeting or phone call. Indeed, ValidSoft has been involved in the detection of deepfake audio since 2011, given deepfake detection is a base requirement in any biometric verification system.

Deepfake Detection Is Essential Across All Channels

Whilst the Juniper Research study was based on e-commerce, i.e., online merchants, e-commerce is certainly not the only attack surface vulnerable to attacks from deepfake. Internet and mobile banking and payments solutions, contact centers, IVRs, and IVAs as well as digital communications such as online meetings have already been or will be targets of deepfake attacks.

ValidSoft recommends the strongest form of customer verification on any channel is biometric verification with deepfake detection integral within it, thereby providing trusted identity assurance impervious to the myriad attack vectors that plague non-biometric solutions, whilst also safeguarding against the threat of deepfakes.

Acknowledging, however, that biometric verification is not universally adopted by all organizations and on all channels, standalone deepfake detection solutions are also required, which can be integrated with any channel or application.

ValidSoft’s standalone deepfake detection solution, Voice Verity®, can acquire snippets of audio from any channel via a simple API, whether it’s from a video, live stream, or even a phone call, and provide an immediate response on the probability of the audio being human or deepfake. So, whilst Juniper is correct in its recommendation to merchants, the potential losses to fraud on other industries and channels, without the same intervention, will be greater again.