FakeWorld 1.0: An Omni-modal Benchmark for Fake Media and Content

Yifeng Gao, Yifan Ding, Li Wang, Feida Huang, Ye Sun, Yixu Wang, Xin Wang, Yutao Wu, Hanxun Huang, Yunhao Feng, Yingshui Tan, Xingjun Ma, Yu-Gang Jiang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:33118-33137, 2026.

Abstract

The rapidly increasing realism of AI-generated media has intensified the spread of deceptive content and undermined public trust. Existing research largely treats this challenge along two separate axes: media authenticity, which assesses whether content is real or machine-generated, and content veracity, which evaluates semantic consistency and factual correctness. This separation overlooks how real-world deception jointly exploits both dimensions. In this work, we present FakeWorld 1.0, an omni-modal benchmark that unifies media authenticity and content veracity within a single evaluation framework. Along the media axis, FakeWorld spans text, audio, image, and video synthesis. Along the content axis, it systematically instantiates cross-modal semantic inconsistencies and factual errors. These two axes are jointly embedded in realistic web-based and streaming-style presentation scenarios, reflecting how multimodal deception is composed, contextualized, and delivered in practice. FakeWorld further provides explainable annotations in the form of per-instance rationales, enabling transparent and evidence-based analysis. Under a unified evaluation protocol, experiments on both open- and closed-source multimodal large language models (MLLMs) reveal fundamental capability limits and demonstrate FakeWorld’s effectiveness in exposing high-fidelity, mixed-source deception. Beyond the benchmark, we introduce OmniChecker, an agentic framwork that performs joint, explainable detection across both axes and produces evidence-backed diagnostic reports. We position FakeWorld 1.0 as a realistic stress test and a practical foundation for advancing scalable, explainable detection of fake multimodal content.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-gao26g, title = {{F}ake{W}orld 1.0: An Omni-modal Benchmark for Fake Media and Content}, author = {Gao, Yifeng and Ding, Yifan and Wang, Li and Huang, Feida and Sun, Ye and Wang, Yixu and Wang, Xin and Wu, Yutao and Huang, Hanxun and Feng, Yunhao and Tan, Yingshui and Ma, Xingjun and Jiang, Yu-Gang}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {33118--33137}, 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/gao26g/gao26g.pdf}, url = {https://proceedings.mlr.press/v306/gao26g.html}, abstract = {The rapidly increasing realism of AI-generated media has intensified the spread of deceptive content and undermined public trust. Existing research largely treats this challenge along two separate axes: media authenticity, which assesses whether content is real or machine-generated, and content veracity, which evaluates semantic consistency and factual correctness. This separation overlooks how real-world deception jointly exploits both dimensions. In this work, we present FakeWorld 1.0, an omni-modal benchmark that unifies media authenticity and content veracity within a single evaluation framework. Along the media axis, FakeWorld spans text, audio, image, and video synthesis. Along the content axis, it systematically instantiates cross-modal semantic inconsistencies and factual errors. These two axes are jointly embedded in realistic web-based and streaming-style presentation scenarios, reflecting how multimodal deception is composed, contextualized, and delivered in practice. FakeWorld further provides explainable annotations in the form of per-instance rationales, enabling transparent and evidence-based analysis. Under a unified evaluation protocol, experiments on both open- and closed-source multimodal large language models (MLLMs) reveal fundamental capability limits and demonstrate FakeWorld’s effectiveness in exposing high-fidelity, mixed-source deception. Beyond the benchmark, we introduce OmniChecker, an agentic framwork that performs joint, explainable detection across both axes and produces evidence-backed diagnostic reports. We position FakeWorld 1.0 as a realistic stress test and a practical foundation for advancing scalable, explainable detection of fake multimodal content.} }
Endnote
%0 Conference Paper %T FakeWorld 1.0: An Omni-modal Benchmark for Fake Media and Content %A Yifeng Gao %A Yifan Ding %A Li Wang %A Feida Huang %A Ye Sun %A Yixu Wang %A Xin Wang %A Yutao Wu %A Hanxun Huang %A Yunhao Feng %A Yingshui Tan %A Xingjun Ma %A Yu-Gang Jiang %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-gao26g %I PMLR %P 33118--33137 %U https://proceedings.mlr.press/v306/gao26g.html %V 306 %X The rapidly increasing realism of AI-generated media has intensified the spread of deceptive content and undermined public trust. Existing research largely treats this challenge along two separate axes: media authenticity, which assesses whether content is real or machine-generated, and content veracity, which evaluates semantic consistency and factual correctness. This separation overlooks how real-world deception jointly exploits both dimensions. In this work, we present FakeWorld 1.0, an omni-modal benchmark that unifies media authenticity and content veracity within a single evaluation framework. Along the media axis, FakeWorld spans text, audio, image, and video synthesis. Along the content axis, it systematically instantiates cross-modal semantic inconsistencies and factual errors. These two axes are jointly embedded in realistic web-based and streaming-style presentation scenarios, reflecting how multimodal deception is composed, contextualized, and delivered in practice. FakeWorld further provides explainable annotations in the form of per-instance rationales, enabling transparent and evidence-based analysis. Under a unified evaluation protocol, experiments on both open- and closed-source multimodal large language models (MLLMs) reveal fundamental capability limits and demonstrate FakeWorld’s effectiveness in exposing high-fidelity, mixed-source deception. Beyond the benchmark, we introduce OmniChecker, an agentic framwork that performs joint, explainable detection across both axes and produces evidence-backed diagnostic reports. We position FakeWorld 1.0 as a realistic stress test and a practical foundation for advancing scalable, explainable detection of fake multimodal content.
APA
Gao, Y., Ding, Y., Wang, L., Huang, F., Sun, Y., Wang, Y., Wang, X., Wu, Y., Huang, H., Feng, Y., Tan, Y., Ma, X. & Jiang, Y.. (2026). FakeWorld 1.0: An Omni-modal Benchmark for Fake Media and Content. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:33118-33137 Available from https://proceedings.mlr.press/v306/gao26g.html.

Related Material