Why Deepfake Fraud Beats Your Workflows, Not Your Technology - with Jon-Rav Shende of Thales Group

In the latest Emerge AI in Business episode, Jon‑Rav Shende of Thales Group warned that synthetic voice fraud is exploding—up 1,300% in 2024 and now averaging seven attempts per day. This surge turns the traditional contact‑center model into a high‑stakes battleground, especially where a caller’s identity, urgency, and a critical business action intersect. Claims processing, payment authorizations, password resets, and executive‑level support become prime targets because agents are measured on speed and customer satisfaction, creating a perfect recipe for deep‑fake manipulation.

Shende emphasized that the threat is less about breaking biometric models and more about hijacking the workflow itself. He advocated a shared‑ownership model where security defines risk thresholds, operations designs the step‑up verification process, and customer experience monitors friction. By classifying actions into low, medium, and high risk—such as account access versus fund transfers—organizations can embed controls that trigger escalation before a fraudulent transaction completes. This collaborative approach satisfies auditors, cyber‑insurers, and regulators while preserving the customer experience.

The episode concluded with a practical four‑step framework: (1) map high‑risk voice journeys, (2) pinpoint decision points where agents must verify identity, (3) establish an evidence chain for audit and forensic needs, and (4) layer AI‑driven risk signals—synthetic‑voice detection, device reputation, and behavioral anomalies—into the workflow. Shende cautioned against tool sprawl, urging leaders to focus on early abnormal‑risk detection rather than accumulating redundant solutions. For senior leaders, the roadmap offers a clear path to transform deep‑fake threats from a reactive nightmare into a manageable, governed risk.