# The Hidden Space of Safety: Understanding Preference-Tuned LLMs in Multilingual context

arXiv: https://arxiv.org/abs/2504.02708 · HTML: https://mlsanity.com/paper/2504.02708/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 21 of its 36 references that are in the map.
- First posted on arXiv 3 Apr 2025.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)
- LLM alignment / RLHF / preference optimisation / [Safety alignment and jailbreak attacks](https://mlsanity.com/alignment/safety-alignment-and-jailbreak-attacks/)

## Builds on (most cited first, 21 of 21)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [A General Theoretical Paradigm to Understand Learning from Human Preferences](https://mlsanity.com/paper/2310.12036/) (2023)
- [Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study](https://mlsanity.com/paper/2305.13860/) (2023)
- [Scalable Extraction of Training Data from (Production) Language Models](https://mlsanity.com/paper/2311.17035/) (2023)
- [Understanding the Effects of RLHF on LLM Generalisation and Diversity](https://mlsanity.com/paper/2310.06452/) (2023)
- [Analyzing Leakage of Personally Identifiable Information in Language Models](https://mlsanity.com/paper/2302.00539/) (2023)
- [LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions](https://mlsanity.com/paper/2304.14402/) (2023)
- [RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs](https://mlsanity.com/paper/2404.08555/) (2024)
- [Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis](https://mlsanity.com/paper/2406.10794/) (2024)
- [Binary Classifier Optimization for Large Language Model Alignment](https://mlsanity.com/paper/2404.04656/) (2024)
- [RLHF Can Speak Many Languages: Unlocking Multilingual Preference Optimization for LLMs](https://mlsanity.com/paper/2407.02552/) (2024)
- [Multilingual != Multicultural: Evaluating Gaps Between Multilingual Capabilities and Cultural Alignment in LLMs](https://mlsanity.com/paper/2502.16534/) (2025)
- [MM-Eval: A Multilingual Meta-Evaluation Benchmark for LLM-as-a-Judge and Reward Models](https://mlsanity.com/paper/2410.17578/) (2024)
- [On Evaluating and Mitigating Gender Biases in Multilingual Settings](https://mlsanity.com/paper/2307.01503/) (2023)
- [Preference Tuning with Human Feedback on Language, Speech, and Vision Tasks: A Survey](https://mlsanity.com/paper/2409.11564/) (2024)
- [LLM Safety Alignment is Divergence Estimation in Disguise](https://mlsanity.com/paper/2502.00657/) (2025)

## Built on by (most cited first, 6 of 6)

- [Refusal Direction is Universal Across Safety-Aligned Languages](https://mlsanity.com/paper/2505.17306/) (2025)
- [MPO: Multilingual Safety Alignment via Reward Gap Optimization](https://mlsanity.com/paper/2505.16869/) (2025)
- [Are LLMs Empathetic to All? Investigating the Influence of Multi-Demographic Personas on a Model's Empathy](https://mlsanity.com/paper/2510.10328/) (2025)
- [LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety](https://mlsanity.com/paper/2604.12710/) (2026)
- [Low-Resource Safety Failures Are Action Failures, Not Representation Failures](https://mlsanity.com/paper/2606.01196/) (2026)
- [LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review](https://mlsanity.com/paper/2608.14626/) (2026)

## Abstract (arXiv)

Alignment tuning has enabled large language models to excel in reasoning, instruction-following, and minimizing harmful generations. However, despite their widespread deployment, these models exhibit a monolingual bias, raising concerns about the effectiveness of alignment across languages. Current alignment methods predominantly focus on English, leaving it unclear how alignment mechanism generalize to multilingual settings. To address this, we conduct a systematic analysis of distributional shifts in the embedding space of LLMs before and after alignment, uncovering its impact on model behavior across diverse languages. We leverage the alignment-induced separation in safety space as a quantitative tool to measure how alignment enforces safety constraints. Our study evaluates seven LLMs using balanced toxicity datasets and parallel text-detoxification benchmarks, revealing substantial disparities in the latent representation space between high-resource and low-resource languages. These findings underscore the need for language-specific fine-tuning to ensure fair, reliable and robust multilingual alignment. Our insights provide a foundation for developing truly safe multilingual LLMs, emphasizing the urgency of addressing alignment gaps in underrepresented languages.