Hierarchical Federated Unlearning for Large Language Models

TL;DR

This paper introduces a scalable federated unlearning method for large language models that effectively manages heterogeneous unlearning requests while preserving model utility and privacy.

Contribution

It proposes a novel hierarchical federated unlearning approach that decouples unlearning and retention, addressing practical challenges in decentralized, sensitive data environments.

Findings

Effectively handles heterogeneous unlearning requests.

Maintains strong language model utility.

Demonstrates scalability and privacy preservation.

Abstract

Large Language Models (LLMs) are increasingly integrated into real-world applications, raising concerns about privacy, security and the need to remove undesirable knowledge. Machine Unlearning has emerged as a promising solution, yet faces two key challenges: (1) practical unlearning needs are often continuous and heterogeneous, and (2) they involve decentralized, sensitive data with asymmetric access. These factors result in inter-domain and intra-domain interference, which further amplifies the dilemma of unbalanced forgetting and retaining performance. In response, we propose a federated unlearning approach for LLMs that is scalable and privacy preserving. Our method decouples unlearning and retention via task-specific adapter learning and employs a hierarchical merging strategy to mitigate conflicting objectives and enables robust, adaptable unlearning updates. Comprehensive experiments on benchmarks of WMDP, MUSE, and TOFU showed that our approach effectively handles heterogeneous unlearning requests while maintaining strong LLM utility compared with baseline methods.