μ2-Bench: A Multilingual Machine Unlearning Benchmark
On the map: LLM safety and security → Machine unlearning
Abstract (arXiv)
Undesired information such as harmful content and private data propagates through Multilingual Large Language Models (LLMs) via direct training and indirect cross-linguistic spread. Multilingual Machine Unlearning (MMU) aims to remove such information, yet its evaluation remains underexplored, leaving unclear whether unlearning truly eliminates target knowledge across all languages. To bridge this gap, we introduce μ2-Bench, an MMU benchmark that simulates the full pipeline of memorization, unlearning, and evaluation across diverse languages. It 1) spans a broad set of languages, 2) evaluates on both training and hold-out languages, and 3) assesses knowledge as dispersed across multiple languages. We show that successful MMU requires methods that reflect multilingual characteristics, and conduct analysis to provide deeper insights into MMU.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 18 of its 24 references that are in the map. First posted on arXiv 17 Sep 2026.
Builds on
18 of 24 references.
Closest ancestors · 11
Same direction, newest first
- 2026Knowledge Beyond Language: Bridging the Gap in Multilingual Machine Unlearning Evaluation14 May
- 2026Layer-Targeted Multilingual Knowledge Erasure in Large Language Models26 Feb
- 2026Evaluating Cross-Lingual Unlearning in Multilingual Language Models10 Jan
- 2026Multilingual Amnesia: On the Transferability of Unlearning in Multilingual LLMs9 Jan
- 2025OpenUnlearning: Accelerating LLM Unlearning via Unified Benchmarking of Methods and Metrics14 Jun
- 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Models8 Jul
- 2024Learn and Unlearn: Addressing Misinformation in Multilingual LLMs19 Jun
- 2024Cross-Lingual Unlearning of Selective Knowledge in Multilingual Language Models18 Jun
- 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
- 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
- 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct
Other references · 7
- 2025Qwen3 Technical Reportcited by 8,820
- 2019How multilingual is Multilingual BERT?cited by 645
- 2020TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languagescited by 488
- 2019Emerging Cross-lingual Structure in Pretrained Language Modelscited by 181
- 2024Multilingual Instruction Tuning With Just a Pinch of Multilingualitycited by 73
- 2024Converging to a Lingua Franca: Evolution of Linguistic Regions and Semantics Alignment in Multilingual Large Language Modelscited by 23
- 2025Rethinking Cross-lingual Alignment: Balancing Transfer and Cultural Erasure in Multilingual LLMscited by 7
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025A Survey on Unlearning in Large Language Models
- 2026Multilingual Unlearning in LLMs: Transfer, Dynamics, and Reversibility
- 2025A Comprehensive Survey of Machine Unlearning Techniques for Large Language Models
- 2026Linguistic Loopholes in LLM Unlearning: From a 174-Language Benchmark to Coverage-Aware Unlearning
- 2024Towards Effective Evaluations and Comparisons for LLM Unlearning Methods
- 2026Beyond Cross-Lingual Transfer: Benchmarking Propagation Boundaries in Multilingual LLM Unlearning
- 2026Unlearning in LLMs: Methods, Evaluation, and Open Challenges
- 2025Existing Large Language Model Unlearning Evaluations Are Inconclusive