Cybersecurity firm Pindrop announced the beta launch of Pindrop Pulse for Meetings, a real-time deepfake detection tool built to detect AI-generated fraud in Zoom, Microsoft Teams, and Webex meetings. With support for more platforms on the way, the move brings Pindrop’s trusted Pulse technology, already used to analyse 130 million phone calls for authenticity, into the complex world of video conferencing.
While companies have spent years fortifying email servers and data networks, video calls have remained an open back door for cybercriminals. Earlier this year, a finance employee at Arup was tricked into wiring $25 million after a deepfake CFO appeared convincingly on a video call. In another alarming case, a North Korean deepfake candidate aced a tech job interview, only to immediately attempt cyber sabotage once the laptop arrived.
Pindrop warns that real-time face swaps, voice cloning, and AI avatars are becoming serious threats.
How does the solution spot deepfakes?
Pindrop Pulse for Meetings works by analysing both audio and video in real time, detecting telltale signs of manipulation that the human eye and ear might miss. It slices video into frames to uncover distorted lighting or shifting facial features, monitors for lip-sync fails between speech and movement, and scrutinises voice patterns for synthetic fingerprints.
It’s a much-needed defence in a world where attackers use four main tricks to pull off video scams: real-time voice modulation, real-time voice conversion, deepfake avatars, and real-time face swaps. The dangers range from financial fraud and corporate espionage to the nightmare scenario of deepfake employees gaining access to sensitive data.
Pindrop, which has spent more than a decade securing banks, insurers, and contact centres, says the new tool is designed to answer a very simple but increasingly crucial question: Is the person on your video call real?