Soteria: Language-Specific Functional Parameter Steering for Multilingual Safety Alignment (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Soteria locates and minimally adjusts the “functional heads” most responsible for harmful content generation in each language. |
| Approach: | Soteria locates and minimally adjusts the "functional heads" responsible for harmful content generation in each language. |
| Outcome: | The proposed approach reduces harmful content generation in languages while preserving model performance. |
Similar Papers
All Languages Matter: On the Multilingual Safety of LLMs (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing safety benchmarks only concern the safety in one language, e.g. the majority language in the pretraining data such as English. |
| Approach: | They propose a prompting method to improve multilingual safety of ChatGPT by enhancing cross-lingual generalization of safety alignment. |
| Outcome: | The proposed method can significantly reduce the ratio of unsafe responses by 42% for non-English queries. |
The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contexts (2024.findings-acl)
Copied to clipboard
Lingfeng Shen, Weiting Tan, Sihao Chen, Yunmo Chen, Jingyu Zhang, Haoran Xu, Boyuan Zheng, Philipp Koehn, Daniel Khashabi
| Challenge: | Recent studies show that malicious prompt instructions could solicit objectionable content from LLMs. |
| Approach: | They compare how state-of-the-art LLMs respond to malicious prompts in different languages . they find that LLM's generate unsafe responses more often when a prompt is written in a lower-resource language . |
| Outcome: | The proposed model can generate unsafe responses more often when a malicious prompt is written in a lower-resource language, and less irrelevant responses when written in lower-source languages. |
The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm (2024.emnlp-main)
Copied to clipboard
null Aakanksha, Arash Ahmadian, Beyza Ermis, Seraphina Goldfarb-Tarrant, Julia Kreutzer, Marzieh Fadaee, Sara Hooker
| Challenge: | Existing approaches to safety alignment focus on homogeneous monolingual settings . preference training and safety measures often overfit to harms common in Western-centric datasets . |
| Approach: | They propose to use human annotated red teaming prompts to identify global and local harms. |
| Outcome: | The proposed approach can address and optimize for a non-homogeneous set of languages and cultural preferences while minimizing both global and local harms. |
Multitask-Bench: Unveiling and Mitigating Safety Gaps in LLMs Fine-tuning (2025.coling-main)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have led to their adoption across a wide range of tasks, ranging from code generation to machine translation and sentiment analysis. |
| Approach: | They propose to fine-tune LLMs on benign (non-harmful) data to ensure safe outputs. |
| Outcome: | The proposed model reduces attack success rates across a range of tasks without compromising its usefulness. |
Multilingual Refusal Alignment for Safer Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are increasingly used globally, but their safety and alignment can vary unpredictably between languages. |
| Approach: | They propose a multilingual refusal alignment dataset to investigate whether alignment transfers cross-lingually and how language consistency is preserved during training. |
| Outcome: | The proposed model can be trained on multilingual datasets without affecting general performance. |
SafeLawBench: Towards Safe Alignment of Large Language Models (2025.findings-acl)
Copied to clipboard
Chuxue Cao, Han Zhu, Jiaming Ji, Qichao Sun, Zhenghao Zhu, Wu Yinyu, Josef Dai, Yaodong Yang, Sirui Han, Yike Guo
| Challenge: | Recent studies indicate that large language models (LLMs) may exhibit risks, including threats to the protection of private data and the generation of hallucinations. |
| Approach: | They propose to evaluate LLMs from a legal perspective using the SafeLawBench benchmark. |
| Outcome: | The proposed framework categorizes safety risks into three levels based on legal standards and includes 24,860 multi-choice questions and 1,106 open-domain question-answering tasks. |
A Survey of Toxicity Mitigation Strategies for Multilingual Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models can reproduce and amplify toxic content, including hate speech, harassment, and bias. |
| Approach: | They propose a comprehensive survey of the many detoxification methods tailored to multilingual LLMs. |
| Outcome: | The proposed methods are based on data filtering, style transfer, expert-based logit steering, retrieval augmentation, and human feedback. |
Safety of Large Language Models Beyond English: A Systematic Literature Review of Risks, Biases, and Safeguards (2026.eacl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) have a growing number of applications that generate harmful, biased, or unsafe content. |
| Approach: | They synthesize findings from recent studies that evaluate their robustness across languages . they highlight gaps in multilingual safety research and recommend future work . |
| Outcome: | The systematic review examines the multilingual safety of large language models in English . it identifies challenges such as dataset availability and evaluation biases . |
Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations (2024.emnlp-main)
Copied to clipboard
| Challenge: | Current alignment methods struggle with dynamic user intentions and complex objectives, making models vulnerable to harmful content. |
| Approach: | They propose a training-free framework that enhances LLM safety across different scenarios. |
| Outcome: | The proposed framework significantly improves safety measures, reduces over-safety, and maintains model utility, outperforming existing methods in ensuring safe content generation. |
Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models achieve effective safety alignment at the time of release, but fine-tuning often compromises safety mechanisms. |
| Approach: | They propose a method that performs safety realignment for large language models . they identify unsafe delta parameters from the fine-tuned models and recalibrate the retained parameters . |
| Outcome: | The proposed method improves safety performance on safety benchmarks and jailbreak attacks while maintaining their performance on downstream tasks. |