arXiv cs.AI· Elif Nebioglu, Emirhan Bilgi\c{c}, Adrian Popescu·· 12 小时前AI 评分58
AI 生成图像检测器依赖全局伪影：INP-X 修复实验揭示检测失效
AI-Generated Image Detectors Exploit Global Artifacts: Evidence from Inpainting Exchange
AI 导读
研究提出 Inpainting Exchange（INP-X）诊断方法，在保留合成内容的同时还原编辑区域外的原始像素，使预训练检测器、商业 API 和去偏方法准确率大幅下降，商业 API 从超过 90% 降至约 55%，去偏方法 DDA 从 97.6% 降至 58.4%。
正文 · 原文
Abstract:Modern diffusion-based inpainting enables realistic, localized image manipulation, raising critical challenges for reliable detection. We show that state-of-the-art AI-generated image detectors largely fail to identify \emph{what} was synthesized and instead exploit a \emph{global} artifact introduced as a side effect of the generation pipeline: latent-diffusion inpainting passes the entire image through a VAE encoder--decoder, inducing a subtle but pervasive spectral shift even in unedited regions. We turn this observation into a simple, model-free diagnostic, \emph{Inpainting Exchange} (INP-X), that restores the original pixels outside the edited region while preserving all synthesized content. To study the effect at scale, we build a benchmark of $30$K matched real, standard-inpainted, and exchanged triplets ($90$K images) spanning four datasets and three inpainters. INP-X requires no model access and is a single compositing step, yet it collapses pretrained detectors, commercial APIs, and bias-aware methods toward chance (commercial APIs: $>$$90\%\!\to\!\sim$$55\%$; the bias-aware DDA: $97.6\%\!\to\!58.4\%$). We give formal propositions linking the effect to high-frequency attenuation from the VAE information bottleneck, and we causally isolate it with a factorial battery and a variance decomposition that attributes $97\%$ of the background spectral shift to the autoencoder round-trip alone. A perceptual study ($n{=}153$) shows the shift is also perceptible to humans, whose accuracy falls from $63.8\%$ to $56.2\%$ ($p{=}0.002$): significantly reduced, though not eliminated. Finally, we show that mechanism-driven mitigations (mixed-distribution and localization-supervised training) improve robustness, but detecting INP-X edits remains an open problem. Our benchmark and code are available at: this https URL
| Comments: | Accepted to WACV 2027. 23 pages, 16 figures, 20 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2602.00192 [cs.CV] |
| (or arXiv:2602.00192v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2602.00192 arXiv-issued DOI via DataCite |
Submission history
From: Emirhan Bilgiç [view email]
[v1]
Fri, 30 Jan 2026 09:14:10 UTC (5,278 KB)
[v2]
Wed, 4 Nov 2026 19:01:47 UTC (5,983 KB)
来源：arXiv cs.AI · arxiv.org