NVIDIA Synthetic Video Detector

Properties

category: detector
maturity: emerging
approach: A NIM (NVIDIA Inference Microservice) within NVIDIA's AI for Media platform that analyzes video frame by frame on NVIDIA GPUs to produce a classifier score for whether a clip contains synthetic content, intended for newsroom and broadcaster review workflows before publication.
access: paid
owns: GPU-accelerated, self-hostable video-authenticity detection microservice distributed to partners such as Wowza, whose Video Intelligence Framework embeds it to bring real-time synthetic-video detection into livestreaming workflows spanning more than 35,000 deployments in over 170 countries, across on-premises, edge, hybrid, and air-gapped environments.
distinctFrom: Unlike Intel FakeCatcher (biological PPG blood-flow signal analysis) or Reality Defender/Sensity AI (cloud-hosted forensic classification services), NVIDIA's detector is a GPU-optimized, self-hostable inference microservice with published compression-robustness benchmarks (92% accuracy uncompressed, dropping to 82% at 50% compression, 22ms per 1080p frame on RTX hardware) built for embedding inside partners' existing video infrastructure like Wowza rather than as a standalone hosted product end users visit directly.
source: NVIDIA Blog: At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI
notes: Announced at SIGGRAPH 2026 on July 20, 2026, roughly two weeks before this lens's 2026-08-05 window start; included as a gap-fill since it was not previously tracked in this catalogue and is materially distinct from existing detector entries, but flagged as pre-window for the assembler's attention. Pricing not disclosed on NVIDIA's site; 'paid' is a provisional best guess for an enterprise NIM microservice distributed via commercial partners like Wowza, not a sourced claim.

Provenance

Source: https://blogs.nvidia.com/blog/siggraph-news-2026/

Properties

category

detector

maturity

emerging

approach

A NIM (NVIDIA Inference Microservice) within NVIDIA's AI for Media platform that analyzes video frame by frame on NVIDIA GPUs to produce a classifier score for whether a clip contains synthetic content, intended for newsroom and broadcaster review workflows before publication.

access

paid

owns

GPU-accelerated, self-hostable video-authenticity detection microservice distributed to partners such as Wowza, whose Video Intelligence Framework embeds it to bring real-time synthetic-video detection into livestreaming workflows spanning more than 35,000 deployments in over 170 countries, across on-premises, edge, hybrid, and air-gapped environments.

distinctFrom

Unlike Intel FakeCatcher (biological PPG blood-flow signal analysis) or Reality Defender/Sensity AI (cloud-hosted forensic classification services), NVIDIA's detector is a GPU-optimized, self-hostable inference microservice with published compression-robustness benchmarks (92% accuracy uncompressed, dropping to 82% at 50% compression, 22ms per 1080p frame on RTX hardware) built for embedding inside partners' existing video infrastructure like Wowza rather than as a standalone hosted product end users visit directly.

source

NVIDIA Blog: At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI

notes

Announced at SIGGRAPH 2026 on July 20, 2026, roughly two weeks before this lens's 2026-08-05 window start; included as a gap-fill since it was not previously tracked in this catalogue and is materially distinct from existing detector entries, but flagged as pre-window for the assembler's attention. Pricing not disclosed on NVIDIA's site; 'paid' is a provisional best guess for an enterprise NIM microservice distributed via commercial partners like Wowza, not a sourced claim.