Zoom expands Pindrop deepfake detection to customer service

Zoom is expanding its collaboration with voice biometrics company Pindrop by adding the company’s identity authentication and deepfake detection to its customer service platform.

​Pindrop has already made its audio deepfake detection tool, Pulse for Meetings, available on Zoom. The product has been named to Time Magazine’s list of the best inventions of 2025.

​The Zoom Contact Center, which allows businesses to manage their customer service, will now also get the firm’s voice authentication feature, Pindrop Passport, and risk analysis tool, Pindrop Protect.

​“By integrating Pindrop’s real-time deepfake detection and voice authentication into our platform, we’re giving organizations stronger tools to help verify callers, reduce fraud risk and maintain confidence in every customer interaction – especially as AI-generated impersonation becomes more sophisticated,” says Kentis Gopalla, head of Zoom CX & Ecosystem.

​Pindrop has already published a case study, showing how Pindrop Pulse helped a large U.S. insurer detect deepfakes and synthetic voice activity in their contact center. The insurer was inspired to beef up its protection after a 2024 incident in Hong Kong, in which an employee of a multinational firm was tricked into transferring US$25 million after a virtual Zoom meeting with deepfakes of senior management.

The U.S.-based firm is also expanding its deepfake detection efforts into the healthcare sector by offering continuous identity verification to HIPAA-regulated environments. The company has published research showing more than half of fraud attempts in healthcare contact centers now involve AI-generated elements.

Voice fraud will only get worse, says Pindrop

In February, Pindrop published a report based on internal data stating that AI-driven fraud surged by more than 1200 percent in 2025. This eye-opening figure is the result of significant improvements in the latency of synthetic voices, enabling fraudsters to impersonate other people more realistically than ever, the firm notes in a newly released analysis.

Pindrop calls this “time-to-first-audio” (TTFA), a delay between the end of a user’s utterance and the moment the system begins responding. Listeners generally treat any TTFA above 1.2 seconds as latency.

In 2025, however, improvements in speech-to-speech reasoning models led to significantly lower latency in speech interaction, with multiple systems operating with a time-to-first-audio of 1.2 seconds or less. Four of these systems arrived in December alone, Sarosh Shahbuddin, Pindrop’s senior director of product, points out in a blog post published last week.

“Seen in that context, the 1,200 percent increase is less an anomaly than an inflection,” writes Shahbuddin. “The technical barriers that once limited sustained, real-time impersonation were removed in 2025.”

There is little reason to expect those barriers to return, meaning that fraudsters’ capabilities in 2026 are higher than they were a year ago and that AI-enabled attacks will continue to grow, he concludes.

Article Topics

AI fraud  |  biometric authentication  |  biometrics  |  deepfake detection  |  generative AI  |  Pindrop  |  voice biometrics  |  Zoom