FairForensics: Seeing Expressions and Parsing Demographics via Vision-Language Modeling for Generalizable Fair Deepfake Detection

2026-08-02

AI Summary

This work addresses the limited generalization of existing deepfake detection methods to unseen forgery types and their fairness disparities across demographic groups. To this end, the authors construct a new deepfake detection benchmark with balanced demographic representation and propose FairForensics, a novel model that disentangles facial expression from demographic attributes during forgery feature extraction. FairForensics integrates an expression encoder, identity-aware image perturbation, demographic-guided language encoding, and a group-aware prototypical objective, all coordinated through a visual–language aligned, demographic-aware fairness regularization mechanism. Evaluated on the proposed benchmark, the method significantly outperforms current approaches, achieving state-of-the-art performance in both cross-forgery generalization and demographic fairness.

Abstract

The challenge of fair deepfake detection (FDD) has attracted increasing attention. Existing fairness-enhanced detectors often suffer from suboptimal generalization to unseen manipulations and fairness across demographic groups. They are typically developed and evaluated on demographically imbalanced distributions, resulting in biased predictions toward minority groups. In this paper, we construct a novel demographically balanced FDD benchmark to train and evaluate the fairness of detectors under both balanced and imbalanced population scenarios. Additionally, we introduce a novel expression and demographic perceptual vision-language model, termed FairForensics, for generalizable fair deepfake detection. FairForensics conducts face forgery generalization enhancement and demographic-aware fairness regularization. During face forgery generalization enhancement, built upon the novel observation of significant distribution differences between pristine and forged expression vectors, we design an expression encoder to capture high-level expression-guided forgery patterns, and an expression-perceptual visual encoder that integrates global appearance and expression forgery features while mitigating identity bias using an identity-aware patch perturbation module. Under demographic-aware fairness regularization, we propose a demographic-guided language encoder to extract population-aware global language embeddings, which boosts the decoupling of forgery features from demographic information via vision-language alignment. We devise a population-aware prototype fairness objective to enforce both inter-class separability and intra-class alignment across demographic subgroups. Extensive experiments on our balanced demographic benchmark show that our method achieves the state-of-the-art in terms of generalization and fairness.

Problem

• fair deepfake detection
• demographic fairness
• generalization
• bias mitigation
• imbalanced datasets

Innovation

• vision-language modeling
• expression-perceptual encoding
• demographic fairness
• deepfake detection
• identity-aware perturbation