SAGA: Source Attribution of Generative AI Videos

1. 1Identify the key characteristics of generative AI models that can be used for source attribution
2. 2Develop a comprehensive framework that can analyze videos and match them to their corresponding generative models
3. 3Train the framework using a large dataset of videos generated by various models
4. 4Evaluate the performance of the framework using metrics such as accuracy and robustness

AI researchers and engineers working on generative models and video analysis can benefit from SAGA to improve the accuracy of source attribution, while product managers and designers can use this technology to develop more effective video verification tools

💡 SAGA provides a way to identify the specific generative model used to create a video, going beyond binary real/fake detection

💡 Introducing SAGA: a framework for attributing generative AI videos to their source models