Retrieval-Driven Training-Free AI-Generated Video Attribution

2026-07-30

AI Summary

This work addresses the urgent need for effective source attribution of increasingly realistic AI-generated videos to support forensic analysis and regulatory oversight. It proposes the first training-free, video-level provenance framework by formulating the problem as an instance retrieval task. The method introduces an end-to-end pipeline for generative fingerprint extraction and cross-frame aggregation, leveraging orthogonal color-space transformations, multi-scale quantized residual generation, and spatiotemporal-semantic feature aggregation. By operating at the video level rather than the image level, this approach overcomes the generalization and scalability limitations of existing techniques. Evaluated on the GenVidBench benchmark, it achieves a Rank-1 accuracy of 20.5% and a mean average precision (mAP) of 16.6%, substantially outperforming current state-of-the-art methods.

Abstract

AI-generated videos are becoming increasingly realistic and difficult to distinguish from authentic ones, which facilitates malicious misuse and poses growing threats to cybersecurity and social governance. Attributing AI-generated videos to their specific generative sources is therefore of critical importance for forensic investigation and legal regulation. However, most existing visual attribution methods focus on images and particularly rely on the image generation model, thereby lacking the ability to generalize to large-scale AI-generated video data. To address these limitations, we introduce an training-free AI-generated video attribution paradigm. Specifically, we formulates AI-generated video attribution as an instance retrieval task, and design a generative fingerprint-based pipeline. This pipeline consists of an adapted orthogonal color transformation, multi-scale quantized residual generation, and temporal-semantic aggregation, progressively capturing and integrating artifacts introduced by generative models across video frames. Extensive experiments on the GenVidBench benchmark demonstrate that our method achieves strong performance in both AI-generated video detection and attribution, outperforming existing state-of-the-art methods with a Rank-1 accuracy of 20.5% and a mean Average Precision of 16.6%. The code is at https://github.com/renxi-seu/Video_Attribution.

Problem

• AI-generated video
• attribution
• forensic investigation
• generative models
• video authenticity

Innovation

• training-free
• video attribution
• generative fingerprint
• instance retrieval
• temporal-semantic aggregation