---
title: "Breaking the Trade-off: Orthogonal Semantic Decoupling for Generalizable and Fair Deepfake Detection"
authors:
  - "Zhongyu Shi"
  - "Siyu Peng"
  - "Yimin Kang"
  - "Ruiyang Xia"
  - "Yizhi Fang"
  - "Weiping Wen"
  - "Zeyu Gu"
  - "Sai Cheng"
doi: "10.24963/ijcai.2026/84"
url: "https://ijcai.org/proceedings/2026/84"
pdf: "https://ijcai.org/proceedings/2026/0084.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "744-752"
track: "Main Track"
keywords:
  - "AI Ethics, Trust, Fairnes: Safety and robustness"
  - "AI Ethics, Trust, Fairnes: Fairness and diversity"
  - "AI Ethics, Trust, Fairnes: Bias"
---

# Breaking the Trade-off: Orthogonal Semantic Decoupling for Generalizable and Fair Deepfake Detection

Zhongyu Shi, Siyu Peng, Yimin Kang, Ruiyang Xia, Yizhi Fang, Weiping Wen, Zeyu Gu, Sai Cheng

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 744-752. DOI: [10.24963/ijcai.2026/84](https://doi.org/10.24963/ijcai.2026/84)

## Abstract

Deepfake detection faces dual challenges in real-world deployment: cross-domain generalization and demographic fairness. Existing approaches often struggle with a trade-off between these goals. Generalization-oriented detectors can over-rely on demographic shortcuts, while fairness constraints tend to steer optimization away from the most discriminative decision boundary. To address this, we propose Orthogonal Semantic Decoupling (OSD), a framework that decouples demographic semantics from forgery cues. Specifically, we perform Singular Value Decomposition on the pretrained weights of a vision-language model, freezing the principal semantic subspace while learning parameter-efficient low-rank experts in the residual subspace. The experts comprise (1) Demographic Semantic Experts, a set of experts specialized via hard sampling and routed based on the similarities between image embeddings and text embeddings of predefined descriptions; and (2) a Universal Forgery Expert, which captures forgery features transferable across domains and demographics. Extensive experiments across multiple benchmarks demonstrate that our approach outperforms state-of-the-art methods in both generalization and fairness, breaking the trade-off. The code is available at https://github.com/sonder-lin/osd-deepfake-detection.