Fair-FLIP: Fair Deepfake Detection with Fairness-Oriented Final Layer Input Prioritising

TL;DR

Fair-FLIP is a novel post-processing method that improves fairness in deepfake detection models by reweighting final layer inputs, reducing demographic biases while preserving detection accuracy.

Contribution

This work introduces Fair-FLIP, a new fairness-oriented post-processing technique that effectively mitigates demographic biases in deepfake detection models without significant accuracy loss.

Findings

Fair-FLIP improves fairness metrics by up to 30%.

Maintains baseline detection accuracy with only 0.25% reduction.

Outperforms existing bias mitigation approaches.

Abstract

Artificial Intelligence-generated content has become increasingly popular, yet its malicious use, particularly the deepfakes, poses a serious threat to public trust and discourse. While deepfake detection methods achieve high predictive performance, they often exhibit biases across demographic attributes such as ethnicity and gender. In this work, we tackle the challenge of fair deepfake detection, aiming to mitigate these biases while maintaining robust detection capabilities. To this end, we propose a novel post-processing approach, referred to as Fairness-Oriented Final Layer Input Prioritising (Fair-FLIP), that reweights a trained model's final-layer inputs to reduce subgroup disparities, prioritising those with low variability while demoting highly variable ones. Experimental results comparing Fair-FLIP to both the baseline (without fairness-oriented de-biasing) and state-of-the-art approaches show that Fair-FLIP can enhance fairness metrics by up to 30% while maintaining baseline accuracy, with only a negligible reduction of 0.25%. Code is available on Github: https://github.com/szandala/fair-deepfake-detection-toolbox