OmniAID: Decoupling Semantics and Artifacts for Universal AI-Generated Image Detection in the Wild

Yuncheng Guo, Junyan Ye, Chenjue Zhang, Hengrui Kang, Haohuan Fu, Conghui He, Weijia Li
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:38614-38641, 2026.

Abstract

A truly universal AI-Generated Image (AIGI) detector must simultaneously generalize across diverse generative models and varied semantic content. Current methods learn a single, entangled forgery representation, conflating content-dependent flaws with content-agnostic artifacts, and are further constrained by outdated benchmarks. We propose OmniAID, a novel framework centered on a decoupled Mixture-of-Experts (MoE) architecture that separates: (1) semantic flaws across distinct content domains via Routable Specialized Semantic Experts, and (2) content-agnostic universal artifacts from content-dependent flaws via a Fixed Universal Artifact Expert. A two-stage training strategy first specializes experts independently with domain-specific hard-sampling, then trains a lightweight gating network for effective input routing. By explicitly decoupling "what is generated” (content-specific flaws) from "how it is generated” (universal artifacts), OmniAID achieves robust generalization. We also introduce Mirage, a large-scale, contemporary dataset comprising a modern training set and a challenging test set. Extensive experiments demonstrate that OmniAID surpasses existing detectors, establishing a new standard for AIGI detection against modern, in-the-wild threats.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-guo26ag, title = {{O}mni{AID}: Decoupling Semantics and Artifacts for Universal {AI}-Generated Image Detection in the Wild}, author = {Guo, Yuncheng and Ye, Junyan and Zhang, Chenjue and Kang, Hengrui and Fu, Haohuan and He, Conghui and Li, Weijia}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {38614--38641}, 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/guo26ag/guo26ag.pdf}, url = {https://proceedings.mlr.press/v306/guo26ag.html}, abstract = {A truly universal AI-Generated Image (AIGI) detector must simultaneously generalize across diverse generative models and varied semantic content. Current methods learn a single, entangled forgery representation, conflating content-dependent flaws with content-agnostic artifacts, and are further constrained by outdated benchmarks. We propose OmniAID, a novel framework centered on a decoupled Mixture-of-Experts (MoE) architecture that separates: (1) semantic flaws across distinct content domains via Routable Specialized Semantic Experts, and (2) content-agnostic universal artifacts from content-dependent flaws via a Fixed Universal Artifact Expert. A two-stage training strategy first specializes experts independently with domain-specific hard-sampling, then trains a lightweight gating network for effective input routing. By explicitly decoupling "what is generated” (content-specific flaws) from "how it is generated” (universal artifacts), OmniAID achieves robust generalization. We also introduce Mirage, a large-scale, contemporary dataset comprising a modern training set and a challenging test set. Extensive experiments demonstrate that OmniAID surpasses existing detectors, establishing a new standard for AIGI detection against modern, in-the-wild threats.} }
Endnote
%0 Conference Paper %T OmniAID: Decoupling Semantics and Artifacts for Universal AI-Generated Image Detection in the Wild %A Yuncheng Guo %A Junyan Ye %A Chenjue Zhang %A Hengrui Kang %A Haohuan Fu %A Conghui He %A Weijia Li %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-guo26ag %I PMLR %P 38614--38641 %U https://proceedings.mlr.press/v306/guo26ag.html %V 306 %X A truly universal AI-Generated Image (AIGI) detector must simultaneously generalize across diverse generative models and varied semantic content. Current methods learn a single, entangled forgery representation, conflating content-dependent flaws with content-agnostic artifacts, and are further constrained by outdated benchmarks. We propose OmniAID, a novel framework centered on a decoupled Mixture-of-Experts (MoE) architecture that separates: (1) semantic flaws across distinct content domains via Routable Specialized Semantic Experts, and (2) content-agnostic universal artifacts from content-dependent flaws via a Fixed Universal Artifact Expert. A two-stage training strategy first specializes experts independently with domain-specific hard-sampling, then trains a lightweight gating network for effective input routing. By explicitly decoupling "what is generated” (content-specific flaws) from "how it is generated” (universal artifacts), OmniAID achieves robust generalization. We also introduce Mirage, a large-scale, contemporary dataset comprising a modern training set and a challenging test set. Extensive experiments demonstrate that OmniAID surpasses existing detectors, establishing a new standard for AIGI detection against modern, in-the-wild threats.
APA
Guo, Y., Ye, J., Zhang, C., Kang, H., Fu, H., He, C. & Li, W.. (2026). OmniAID: Decoupling Semantics and Artifacts for Universal AI-Generated Image Detection in the Wild. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:38614-38641 Available from https://proceedings.mlr.press/v306/guo26ag.html.

Related Material