New tool detects AI-generated videos with 93.7% accuracy
What is the purpose of the DIVID tool?
The DIVID tool, developed by Columbia Engineering researchers, is designed to detect AI-generated videos with high accuracy. It expands on previous work, Raidar, which detects AI-generated text. The tool is needed as AI-generated videos have become increasingly realistic, making it difficult for humans and existing detection systems to distinguish between real and fake videos.
How does DIVID differ from previous AI detection methods?
DIVID, a new AI-generated video detection tool, improves upon previous methods by analyzing diffusion-generated videos and using DIRE values to identify AI-generated content. Unlike older methods that focused on detecting generative videos from GANs, DIVID is designed to handle the increasing sophistication of AI-generated videos, such as those created by diffusion models like Sora by OpenAI.
How does the DIRE technique function in detecting AI videos?
The DIRE (DIffusion Reconstruction Error) technique detects AI-generated videos by measuring the difference between an input image and the corresponding output image reconstructed by a pretrained diffusion model5. This helps identify videos generated by diffusion models, which are becoming increasingly sophisticated and difficult to distinguish from real videos.