A Difference-in-Difference Approach to Detecting AI-Generated Images

Xinyi Qi, Kai Ye, Chengchun Shi, Ying Yang, Hongyi Zhou, Jin Zhu · 2026

Read A Difference-in-Difference Approach to Detecting AI-Generated Images by Xinyi Qi, Kai Ye, Chengchun Shi, Ying Yang, Hongyi Zhou, Jin Zhu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-23732.txt.

Description

Xinyi Qi and five co-authors propose a gap-in-gap method for spotting AI-made images. Spread models produce AI-made images almost twin-like to real ones, raising misuse concerns and large spotting hurdles. Many current spotters rely on rebuild error, the gap between an input image and its rebuilt version, to tell real from fake images.

Those spotters weaken as modern AI-made images grow more and more similar to real ones. The new method instead finds the gap in rebuild error, a second-order gap, for variance cut and better spotting scores. Wide tests show strong reach, giving reliable spotting of AI-made images in the era of making AI.

Catalog record

CREATORXinyi Qi, Kai Ye, Chengchun Shi, Ying Yang, Hongyi Zhou, Jin Zhu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-23732

Browse the full library · About the Tower of Babel