As the advancement of generative AI technology enables the creation of convincing deepfakes of a person’s likeness and voice, fraudsters are beginning to deploy such deepfakes to bypass KYC security verification, impersonate executives, and manipulate financial institutions and other enterprises into initiating high-risk actions. 

In the first quarter of 2024, the use of conferencing apps like Zoom as the second step in multi-channel fraud attacks has increased by 33.3%. AI-fueled cybercrime and deepfakes are the most pressing challenges that businesses expect to encounter in the next 2-3 years.

The rise in video deepfakes has already resulted in massive losses for companies. In a well-publicized case, a worker for the engineering company Arup was manipulated by a fake video call with the company’s CFO into transferring $25 million into a fraudster’s account. As cybercriminals continue to test companies’ defenses with new AI-fueled scams, Deloitte estimates that generative AI will enable $40 billion in fraud losses in the United States by 2027. 43% of surveyed finance professionals have fallen victim to a deepfake scamming attack. Yet, 87% of the same professionals admit that they would initiate a transaction if they were “called” by their CEO or CFO to do so without delays.

Real-Time Video Deepfake Detection

To support companies as they face this growing threat, the Reality Defender team is proud to announce the release of our enhanced Real-Time Video Deepfake Detection. 

This cutting-edge Reality Defender plugin integrates directly with web conferencing vendors as a native app, providing a convenient, powerful cybersecurity tool running at the start of your calls. 

Real-Time Video Deepfake Detection adds a crucial layer to internal security and external KYC/AML client verification, securing peace of mind for executives, workers, and customers, who can rest assured they are looking at a real person, not an AI-generated impersonation with malicious intent. The ease of integration and use ensures that the plugin is ready to go on day one, providing a premium cybersecurity solution that is built to scale, platform-agnostic, and fueled by Reality Defender’s award-winning detection models.

How it Works

After a simple addition of the plugin into the conferencing platform, the Reality Defender interface appears directly within the application. When a call is initiated, the Reality Defender multi-model detection suite joins the call as a bot, displaying a side panel visible only to the user who initiated the call. 

As the call proceeds, Reality Defender’s video detection models begin to analyze the faces on the screen in real time, populating the side panel with results and providing a score that reveals the likelihood of AI manipulation within the call. If deepfake technology is being used to alter a participant’s likeness and spoof KYC verification flow or impersonate a trusted executive, customer, or employee, the user is notified right away with an actionable report.

The individual video streams are captured in the original resolution and with raw image information for each participant. For convenience, the Reality Defender plugin bot can join multiple meetings, and the user interface offers a set of tools to manage meetings and disconnect the detection bots at will. The interface also enables coordination of multiple video flows and aggregation of results per each meeting and participant. The plugin is now available for Zoom, with updates supporting other conferencing applications coming soon.

Real-time video deepfakes are here to stay, and are being used for fraud and nefarious purposes more at an unfathomable pace. Which is why Reality Defender can now detect these deepfakes every time you join a meeting, backed by our famous multi-model approach.

Click here to sign up for the Real-Time Video Deepfake Detection private beta and start using deepfake detection in your web conferencing platforms today.