Nvidia took advantage of an event that took place in Los Angeles to unveil the Synthetic Video Detector, a new tool capable of identifying videos created with Artificial Intelligence in less than 1 second.
© Getty Images
21/07/2026
by
A as Artificial Intelligence companies work on increasingly powerful models – capable of generating more realistic images and videos than ever before – we have seen appeals and pressures to create tools capable of identifying these contents.
It is in this context that Nvidia took advantage of the SIGGRAPH 2026 event in Los Angeles, USA, to unveil the Synthetic Video Detector.
This new tool was created to quickly identify videos created with Artificial Intelligence, with Nvidia indicating that it is capable of processing a 1080p resolution video in only 22 milliseconds – which makes this tool a viable way to ascertain the veracity of a video in a real situation.
In addition to speed, Nvidia also highlights the accuracy of the Synthetic Video Detector. The company notes that, in non-compressed videos, it was able to achieve an accuracy of 92%, and in compressed videos up to 15% this accuracy falls to 87% and 82% in compressed videos up to 50%.
It is important to note that the effectiveness of this detection tool with compressed videos is important to identify content generated with Artificial Intelligence on social networks.
It is worth remembering that videos posted on social networks such as Instagram, TikTok and YouTube are always subject to some compression, so artificial intelligence detection tools will always have to be prepared to deal with this issue.
How does the digital Trends website, Nvidia notes that, although the Synthetic Video Detector does not solve the challenge of identifying Artificial Intelligence videos at 100%, it may be a valuable help for means of verification as technology continues to evolve.
The idea is to provide the Synthetic Video Detector in about 170 countries in the coming months.

