GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuning

• URL: http://arxiv.org/abs/2603.10243v1
• Date: Tue, 10 Mar 2026 21:40:09 GMT
• Title: GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuning
• Abstract summary: Safety alignment of large language models can be easily compromised by seemingly non-adversarial fine-tuning.<n>Inspired by generative replay in continual learning, we propose Generative Replay for Safety Alignment Preservation (GR-SAP)<n>Experiments show GR-SAP substantially mitigates fine-tuning-induced safety degradation while maintaining comparable downstream performance.
• Score: 15.234182153613348
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent studies show that the safety alignment of large language models (LLMs) can be easily compromised even by seemingly non-adversarial fine-tuning. To preserve safety alignment during fine-tuning, a widely used strategy is to jointly optimize safety and task objectives by mixing in the original alignment data, which is typically inaccessible even for open-weight LLMs. Inspired by generative replay in continual learning, we propose Generative Replay for Safety Alignment Preservation (GR-SAP), a unified framework that synthesizes domain-specific alignment data from LLMs and integrate them during downstream adaption to preserve safety alignment. Theoretical and empirical analyses demonstrate this synthetic data serves as a reliable proxy for the original alignment data. Experiments across various models and downstream tasks show that GR-SAP substantially mitigates fine-tuning-induced safety degradation while maintaining comparable downstream performance. Our code is available at https://github.com/chili-lab/gr-sap.

Related papers

• Safeguarding LLM Fine-tuning via Push-Pull Distributional Alignment [45.772620696660034]
  Safety Optimal Transport (SOT) is a novel framework that reframes safe fine-tuning from an instance-level filtering challenge to a distribution-level alignment task grounded in Optimal Transport (OT)<n>SOT prioritizes sample importance by actively pulling the downstream distribution towards a trusted safe anchor while simultaneously pushing it away from a general harmful reference.<n>Experiments across diverse model families and domains demonstrate that SOT significantly enhances model safety while maintaining competitive downstream performance.
  arXiv  Detail & Related papers  (2026-01-12T04:48:02Z)
• AlignGuard-LoRA: Alignment-Preserving Fine-Tuning via Fisher-Guided Decomposition and Riemannian-Geodesic Collision Regularization [6.5225344327304535]
  Low-rank adaptation (LoRA) has become a standard tool for efficiently fine-tuning large language models.<n>LoRA updates can induce alignment drift, weakening safety and behavioral constraints.<n>We propose AlignGuard-LoRA, a principled framework for preserving alignment during finetuning.
  arXiv  Detail & Related papers  (2025-08-04T05:45:24Z)
• Layer-Aware Representation Filtering: Purifying Finetuning Data to Preserve LLM Safety Alignment [24.364891513019444]
  In this paper, we show that fine-tuning datasets often contain samples with safety-degrading features that are not easily identifiable on the surface.<n>We propose LARF, a Layer-Aware Representation Filtering method.<n> Experimental results demonstrate that LARF can effectively identify benign data with safety-degrading features.
  arXiv  Detail & Related papers  (2025-07-24T17:59:24Z)
• Safe Pruning LoRA: Robust Distance-Guided Pruning for Safety Alignment in Adaptation of LLMs [4.580092836731863]
  Fine-tuning Large Language Models (LLMs) with Low-Rank Adaptation (LoRA) enhances adaptability while reducing computational costs.<n>Existing safety alignment methods struggle to capture complex parameter shifts, leading to suboptimal safety-utility trade-offs.<n>We propose Safe Pruning LoRA (SPLoRA), a novel pruning-based approach that selectively removes LoRA layers that weaken safety alignment.
  arXiv  Detail & Related papers  (2025-06-21T14:59:54Z)
• Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets [64.96967819446553]
  This paper investigates the degradation of safety guardrails through the lens of representation similarity between upstream alignment datasets and downstream fine-tuning tasks.<n>High similarity between these datasets significantly weakens safety guardrails, making models more susceptible to jailbreaks.<n>Low similarity between these two types of datasets yields substantially more robust models and thus reduces harmfulness score by up to 10.33%.
  arXiv  Detail & Related papers  (2025-06-05T17:59:55Z)
• Shape it Up! Restoring LLM Safety during Finetuning [65.75757313781104]
  Finetuning large language models (LLMs) enables user-specific customization but introduces critical safety risks.<n>We propose dynamic safety shaping (DSS), a framework that uses fine-grained safety signals to reinforce learning from safe segments of a response while suppressing unsafe content.<n>We present STAR-DSS, guided by STAR scores, that robustly mitigates finetuning risks and delivers substantial safety improvements across diverse threats, datasets, and model families.
  arXiv  Detail & Related papers  (2025-05-22T18:05:16Z)
• Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models? [83.53005932513155]
  Multi-modal large language models (MLLMs) have made significant progress, yet their safety alignment remains limited.<n>We propose finetuning MLLMs on a small set of benign instruct-following data with responses replaced by simple, clear rejection sentences.
  arXiv  Detail & Related papers  (2025-04-14T09:03:51Z)
• Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models [30.93821289892195]
  We propose IRR (Identify, Remove, and Recalibrate for Safety Realignment) that performs safety realignment for LLMs.<n>The core of IRR is to identify and remove unsafe delta parameters from the fine-tuned models, while recalibrating the retained ones.<n>Our results demonstrate that IRR significantly enhances the safety performance of fine-tuned models on safety benchmarks, such as harmful queries and jailbreak attacks.
  arXiv  Detail & Related papers  (2024-12-15T03:58:38Z)
• Fake Alignment: Are LLMs Really Aligned Well? [91.26543768665778]
  This study investigates the substantial discrepancy in performance between multiple-choice questions and open-ended questions. Inspired by research on jailbreak attack patterns, we argue this is caused by mismatched generalization.
  arXiv  Detail & Related papers  (2023-11-10T08:01:23Z)
• Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To! [88.90694413503614]
  We find that the safety alignment of LLMs can be compromised by fine-tuning. We jailbreak GPT-3.5 Turbo's safety guardrails by fine-tuning it on only 10 such examples. We advocate for further research efforts toward reinforcing safety protocols for the custom fine-tuning of aligned LLMs.
  arXiv  Detail & Related papers  (2023-10-05T17:12:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.