Russia Has Developed a High-Precision Deepfake Detector for Social Networks
Russia has developed a high‑precision deepfake detector for social networks, a tool designed to counter digital fraud and large‑scale manipulation as AI‑generated content becomes increasingly convincing and harder to identify.
A New AI Shield Against Deepfakes
Russian company MTS Web Services has unveiled an advanced deepfake‑detection system capable of identifying AI‑generated video content with up to 98% accuracy. According to the developers, the tool analyzes both visual and audio components to determine whether a clip has been manipulated.
The system uses a VisionLabs detector for video verification, while the audio track is processed by the proprietary MWS AI model. The model is trained first on real human speech and then on synthetic recordings, allowing it to recognize the distinct markers of artificial voices.
The technology is already undergoing testing on the MTS Link video‑conferencing platform, in the Defender anti‑fraud calling service, within government digital services, and across several banks.
Addressing Fraud and Information Manipulation
The rise of highly realistic deepfakes poses significant risks. Fake videos and audio messages are increasingly used by malicious actors to deceive individuals, impersonate public figures, or spread panic through fabricated political statements or staged disasters. Without strong detection mechanisms, distinguishing AI‑generated content from real footage becomes nearly impossible.
The MTS system tackles this challenge head‑on — and is designed for rapid evolution. MWS CEO Pavel Voronin noted that training the system to detect deepfakes produced by the newest content‑generation models took only two weeks.
A Tool Built for the Future
As deepfake technology advances, the MTS detector is positioned as a critical safeguard for digital ecosystems, helping protect citizens, institutions, and media spaces from misinformation and large‑scale manipulation.
