Bias Redistribution in Visual Machine Unlearning: Does Forgetting One Group Harm Another?

TL;DR

This paper examines how machine unlearning affects bias redistribution in CLIP models, revealing that forgetting one demographic group often shifts bias to others, especially along gender lines.

Contribution

It uncovers bias redistribution phenomena in CLIP models during unlearning and evaluates methods that mitigate but do not fully prevent bias transfer.

Findings

Unlearning redistributes bias mainly along gender boundaries.

Removing Young Female transfers performance to Old Female.

Refusal Vector reduces redistribution but degrades overall performance.

Abstract

Machine unlearning enables models to selectively forget training data, driven by privacy regulations such as GDPR and CCPA. However, its fairness implications remain underexplored: when a model forgets a demographic group, does it neutralize that concept or redistribute it to correlated groups, potentially amplifying bias? We investigate this bias redistribution phenomenon on CelebA using CLIP models (ViT/B-32, ViT-L/14, ViT-B/16) under a zero-shot classification setting across intersectional groups defined by age and gender. We evaluate three unlearning methods, Prompt Erasure, Prompt Reweighting, and Refusal Vector using per-group accuracy shifts, demographic parity gaps, and a redistribution score. Our results show that unlearning does not eliminate bias but redistributes it primarily along gender rather than age boundaries. In particular, removing the dominant Young Female group consistently transfers performance to Old Female across all model scales, revealing a gender-dominant structure in CLIP's embedding space. While the Refusal Vector method reduces redistribution, it fails to achieve complete forgetting and significantly degrades retained performance. These findings highlight a fundamental limitation of current unlearning methods: without accounting for embedding geometry, they risk amplifying bias in retained groups.