Where Detectors Fail: Probing Generative Space for Generalizable AI-Generated Image Detection

Zijie Cao, Weijie Tu, Yao Xiao, Weijian Deng, Weiyan Chen, Liang Lin, Pengxu Wei
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:11477-11495, 2026.

Abstract

Detecting AI-generated images (AIGI) remains challenging because detectors often fail to generalize to unseen generators. Although existing methods are trained on large datasets, their performance still degrades when generation settings change, indicating that data scale alone is insufficient and that limited coverage of generative variations during training is a key factor. Studies on generative model editing show that small changes in internal representations can produce diverse and meaningful image variations, many of which are not explored under standard sampling. Leveraging this insight, we propose PROBE (Probing Robustness via Boundary Exploration), a framework that improves detector generalization by actively exploring challenging regions of the generative process. Instead of treating the generator as a fixed data source, PROBE uses the detector as a critic to steer the generator through manifold-level modifications, producing realistic samples that are difficult to classify. These samples expose failure cases that are uncommon under standard data sampling strategies and are used to refine the detector. Experimental results across multiple benchmarks indicate that PROBE enhances generalization to unseen generators, resulting in more generalizable AIGI detection performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-cao26s, title = {Where Detectors Fail: Probing Generative Space for Generalizable {AI}-Generated Image Detection}, author = {Cao, Zijie and Tu, Weijie and Xiao, Yao and Deng, Weijian and Chen, Weiyan and Lin, Liang and Wei, Pengxu}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {11477--11495}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/cao26s/cao26s.pdf}, url = {https://proceedings.mlr.press/v306/cao26s.html}, abstract = {Detecting AI-generated images (AIGI) remains challenging because detectors often fail to generalize to unseen generators. Although existing methods are trained on large datasets, their performance still degrades when generation settings change, indicating that data scale alone is insufficient and that limited coverage of generative variations during training is a key factor. Studies on generative model editing show that small changes in internal representations can produce diverse and meaningful image variations, many of which are not explored under standard sampling. Leveraging this insight, we propose PROBE (Probing Robustness via Boundary Exploration), a framework that improves detector generalization by actively exploring challenging regions of the generative process. Instead of treating the generator as a fixed data source, PROBE uses the detector as a critic to steer the generator through manifold-level modifications, producing realistic samples that are difficult to classify. These samples expose failure cases that are uncommon under standard data sampling strategies and are used to refine the detector. Experimental results across multiple benchmarks indicate that PROBE enhances generalization to unseen generators, resulting in more generalizable AIGI detection performance.} }
Endnote
%0 Conference Paper %T Where Detectors Fail: Probing Generative Space for Generalizable AI-Generated Image Detection %A Zijie Cao %A Weijie Tu %A Yao Xiao %A Weijian Deng %A Weiyan Chen %A Liang Lin %A Pengxu Wei %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-cao26s %I PMLR %P 11477--11495 %U https://proceedings.mlr.press/v306/cao26s.html %V 306 %X Detecting AI-generated images (AIGI) remains challenging because detectors often fail to generalize to unseen generators. Although existing methods are trained on large datasets, their performance still degrades when generation settings change, indicating that data scale alone is insufficient and that limited coverage of generative variations during training is a key factor. Studies on generative model editing show that small changes in internal representations can produce diverse and meaningful image variations, many of which are not explored under standard sampling. Leveraging this insight, we propose PROBE (Probing Robustness via Boundary Exploration), a framework that improves detector generalization by actively exploring challenging regions of the generative process. Instead of treating the generator as a fixed data source, PROBE uses the detector as a critic to steer the generator through manifold-level modifications, producing realistic samples that are difficult to classify. These samples expose failure cases that are uncommon under standard data sampling strategies and are used to refine the detector. Experimental results across multiple benchmarks indicate that PROBE enhances generalization to unseen generators, resulting in more generalizable AIGI detection performance.
APA
Cao, Z., Tu, W., Xiao, Y., Deng, W., Chen, W., Lin, L. & Wei, P.. (2026). Where Detectors Fail: Probing Generative Space for Generalizable AI-Generated Image Detection. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:11477-11495 Available from https://proceedings.mlr.press/v306/cao26s.html.

Related Material