Safeguarding LLM Fine-tuning via Push-Pull Distributional Alignment (2026.acl-long)
Copied to clipboard
| Challenge: | Existing safety defenses for large language models fail to explicitly repel harmful patterns . Optimal transport (SOT) allows for safe fine-tuning without sacrificing safety . |
| Approach: | They propose a framework that reframes safe fine-tuning from instance-level filtering challenge to distribution-level alignment task grounded in Optimal Transport. |
| Outcome: | a new framework improves safety of large language models while maintaining competitive performance . the proposed framework reduces the risk of errors and improves model performance compared to baselines . |
Similar Papers
Safeguard Fine-Tuned LLMs Through Pre- and Post-Tuning Model Merging (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Fine-tuning large language models for downstream tasks often leads to catastrophic forgetting, notably degrading the safety of original alignments. |
| Approach: | They propose to merge the weights of pre- and post-fine-tuned models to improve safety while enhancing performance. |
| Outcome: | Experiments across different downstream tasks and models validate the method’s practicality and effectiveness. |
The Art of (Mis)alignment: How Fine-Tuning Methods Effectively Misalign and Realign LLMs in Post-Training (2026.findings-acl)
Copied to clipboard
Rui Zhang, Hongwei Li, Yun Shen, Xinyue Shen, Wenbo Jiang, Guowen Xu, Yang Liu, Michael Backes, Yang Zhang
| Challenge: | Misaligned large language models can magnify harm by exploiting them to undermine safety . et al., 2022b; Bai e.t., 2023): misalignment, realignment and model-specific resistance are important . |
| Approach: | They evaluate four methods to identify a mechanism asymmetry between attack and defense . they find that ORPO is most effective for misalignment, but DPO excels in realignment . |
| Outcome: | The proposed methods show a mechanism asymmetry between attack and defense . the proposed methods excel in realignment, but at the expense of model utility . |
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. |
SafeMERGE: Preserving Safety Alignment in Fine-Tuned Large Language Models via Selective Layer-Wise Model Merging (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent studies show that fine-tuning can erode safety alignment, causing LLMs to respond to harmful or unethical prompts. |
| Approach: | They propose a lightweight framework that restores safety while maintaining downstream performance. |
| Outcome: | The proposed framework reduces harmful outputs compared to other defenses, with negligible impact on utility. |
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. |
OASIS: Mitigating Harmful Fine-tuning Attacks on LLMs via Orthogonal and Adaptive Safety Alignment Strategy (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods to decouple safety enforcement from harmful feature acquisition rely on perturbation directions that conflict with harmful gradients . harmful fine-tuning attacks pose a significant challenge for service providers aiming to uphold rigorous safety standards. |
| Approach: | They propose an orthogonal and ad hoc safety alignment strategy to decouple safety enforcement from harmful feature acquisition. |
| Outcome: | Experiments on four large language models show that OASIS reduces the Harmful Score by 60% compared to baselines while maintaining stable task utility. |
Ensuring Safe and High-Quality Outputs: A Guideline Library Approach for Language Models (2024.naacl-long)
Copied to clipboard
Yi Luo, Zhenghao Lin, YuHao Zhang, Jiashuo Sun, Chen Lin, Chengjin Xu, Xiangdong Su, Yelong Shen, Jian Guo, Yeyun Gong
| Challenge: | Guide-Align is a guideline-oriented approach to augment the safety and quality of Large Language Models. |
| Approach: | They propose a guideline-oriented method to augment the safety and quality of large language models. |
| Outcome: | The proposed method outperforms existing methods on three benchmarks and shows significant improvements in security and quality. |
Towards Identification and Intervention of Safety-Critical Parameters in Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing safety-related methodologies for large language models are lacking . despite advances in safety alignment techniques, safeguarding LLMs during adaptation to various tasks remains a challenge. |
| Approach: | They propose a framework to quantify how different parameters affect LLM safety . they propose two targeted intervention paradigms for safety enhancement and preservation . |
| Outcome: | The proposed framework reveals safety-critical patterns across different LLM architectures. |
TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis (2025.acl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in natural language processing tasks but are vulnerable to harmful content and being exploited for malicious purposes. |
| Approach: | They propose a framework to measure the risk coverage of alignment datasets across three dimensions: Lexical Diversity, Malicious Intent, and Jailbreak Tactics. |
| Outcome: | The proposed framework measures risk coverage across Lexical Diversity, Malicious Intent, and Jailbreak Tactics. |
Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets (2026.acl-long)
Copied to clipboard
| Challenge: | Existing mitigation strategies focus on reactively addressing jailbreak incidents after safety guardrails have been compromised. |
| Approach: | They investigate the degradation of safety guardrails through the lens of representation similarity between upstream alignment datasets and downstream fine-tuning tasks. |
| Outcome: | The proposed model reduces harmfulness score by 10.33% when compared to baseline models. |