NVIDIA has released a new software tool called the Synthetic Video Detector to help organizations identify AI-generated video content. This release matters because broadcasters, financial institutions, and government agencies face strict security requirements and need reliable ways to verify media authenticity. The tool addresses the growing risk of synthetic media by providing a technical solution for content validation.
Tool analyzes video frame-by-frame to classify synthetic content
The product functions as a NIM microservice that analyzes video frame-by-frame to classify synthetic content. NVIDIA introduced the tool at SIGGRAPH in Los Angeles as part of its broader software ecosystem. The system is designed to integrate with existing workflows for organizations that require operational control and data residency compliance.
The detector achieves 92% accuracy on uncompressed video and completes the analysis in less than a second. Processing speed reaches 22 milliseconds for 1080p video on RTX systems and around 30 milliseconds on Ada Lovelace NVIDIA L40 enterprise-grade data center GPUs. The tool is built to handle high-volume processing while maintaining low latency for real-time or near-real-time verification.
Detection accuracy drops slightly when video compression is applied, reaching 85% for 15% compression and 82% for 50% compression. This performance profile allows the system to function effectively even when media has been modified for distribution. The technology has already been deployed by Wowza in over 35,000 locations across more than 170 countries.
NVIDIA states that this scale is critical for organizations exposed to synthetic media risk, including critical infrastructure operators. The company emphasizes that strict requirements around data residency and security drive the need for such distributed detection capabilities. The Synthetic Video Detector is now available as a functional component of the NVIDIA software suite.
Source: TweakTown
Discussion
0 comments