Effective protection against deepfakes
The total number of deepfake verifications based on voice and face is forecast to more than double within two years, from over 6 billion in 2026 to 12.2 billion in 2028, and annual market revenue is forecast to increase from $3 billion to $6 billion, according to the latest report, "2026 Deepfake Fraud Detection Market Report: Securing Identity in the AI Era," from Biometric Update.
The report confirms that deepfake detection will increasingly be integrated into broader identity platforms rather than implemented as a standalone feature, reflecting a market shift toward multi-layered, AI-based fraud protection.
Examples of such safeguards include:
- Voice-based deepfake audio detection models that process raw audio to determine the likelihood of fraud (e.g., the Velma-2 algorithm from Modulate).
- Neural audio watermarking, which is a hidden signal added to an audio file by a machine learning model trained to seamlessly but reliably embed signals in audio waves (e.g., the Detect algorithm from Resemble).
- Real-time AI fraud detection in corporate meetings on Zoom, Google Meet, and Microsoft Teams (Polygraf's Meeting Guard tool).
Meeting Guard is a visible participant in virtual meetings that acts as a secure compliance assistant, detecting AI-generated content, verifying voices for deepfake threats, and generating secure notes.
more in the article
https://www.biometricupdate.com/202607/deepfake-threat-pushes-detection-beyond-single-biometric-signals