Auditing Fairness-Privacy Trade-offs: Subpopulation-Level Effects of Fairness-Enhancing Algorithms

2026-07-16

AI Summary

This study investigates how fairness-enhancing algorithms affect the privacy leakage risks faced by distinct subgroups, with a focus on the trade-off between privacy and fairness under membership inference attacks. To this end, it introduces the first subgroup-aware joint audit framework for fairness and privacy, integrating an improved likelihood ratio attack (LiRA), differential privacy mechanisms, and multiple fairness interventions, and conducts systematic evaluations across diverse model architectures. The findings reveal that the impact of fairness interventions on privacy risk is non-uniform, varying with model architecture, subgroup size, and the specific fairness strategy employed. Moreover, the utility gains and privacy costs of differential privacy are unevenly distributed across subgroups, uncovering critical disparities obscured by aggregate-level assessments.

Abstract

Machine learning (ML) models deployed in sensitive domains such as healthcare, law enforcement, and finance must satisfy not only utility requirements but also fairness and privacy guarantees. While prior work has largely examined how privacy-preserving techniques affect fairness, the inverse question-how fairness-enhancing algorithms influence privacy leakage-remains underexplored. We present the first comprehensive study of how fairness interventions affect membership inference privacy risks at the subpopulation level. By adapting the Likelihood Ratio Attack (LiRA) for subgroup auditing, we uncover privacy disparities that aggregate evaluations obscure. We further analyze how Differential Privacy (DP) interacts with fairness-enhancing methods across different categories, showing that DP's privacy benefits and utility costs are unevenly distributed across subpopulations. Our results demonstrate that fairness interventions do not uniformly increase privacy risk; their impact depends on model architecture, subgroup size, and mitigation strategy. These findings reveal that fairness, privacy, and utility must be jointly evaluated at the subpopulation level, and we introduce the first unified empirical framework to support such auditing in practice.

Problem

• fairness-privacy trade-offs
• subpopulation-level privacy
• membership inference
• fairness-enhancing algorithms
• differential privacy

Innovation

• fairness-privacy trade-offs
• subpopulation-level auditing
• membership inference attack
• differential privacy
• Likelihood Ratio Attack