When Eyes Betray AI: Social Gaze Consistency as a Semantic Cue for AI-Generated Image Detection

Jihyeon Kim, Sohee Kim, Soosan Lee, Souhwan Jung, James Matthew Rehg, Hyesong Choi · 2026

Read When Eyes Betray AI: Social Gaze Consistency as a Semantic Cue for AI-Generated Image Detection by Jihyeon Kim, Sohee Kim, Soosan Lee, Souhwan Jung, James Matthew Rehg, Hyesong Choi (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-2605-27348.txt.

Description

Jihyeon Kim and colleagues detect fake images from gaze: whether the eyes, head, and pupils of people in a scene agree with each other. Their supervision holds a fixed five-block reasoning skeleton across many captions, blocking shortcuts from paired edits. The same training lifts a vision-language detector and a vision-only detector, with real and fake recall rising together.

Catalog record

CREATORJihyeon Kim, Sohee Kim, Soosan Lee, Souhwan Jung, James Matthew Rehg, Hyesong Choi
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75036
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.27348

Browse the full library · About the Tower of Babel