AI media detectors with AUC near 0.99 in clean conditions suffer major accuracy and calibration drops under platform transforms and visually constrained adversarial attacks.
Towards uni- versal fake image detectors that generalize across generative models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The Deployment Gap in AI Media Detection: Platform-Aware and Visually Constrained Adversarial Evaluation
AI media detectors with AUC near 0.99 in clean conditions suffer major accuracy and calibration drops under platform transforms and visually constrained adversarial attacks.