DataShield: Safety-degrading Data Filtering for LLM Benign Instruction Fine-Tuning

Computer Science > Cryptography and Security [Submitted on 29 May 2026] DataShield: Safety-degrading Data Filtering for LLM Benign Instruction Fine-Tuning Junbo Zhang, Qianli Zhou, Xinyang Deng, Wen Jiang, Jie Pan, Jinbiao Zhu Large language models (LLMs) suffer from degraded safety capabilities even when fine-tuned with benign datasets. However, existing methods for identifying safety-degrading samples in benign datasets suffer from high computational costs and significant noise issues. In this paper, we propose DataShield to efficiently and effectively identify potential safety-degrading samples. Our key intuition is based on the observation that benign fine-tuning increases the overall response compliance of LLMs. DataShield's key technical insight is to quantify each sample's contribution to the model's compliance behavior as its safety degradation score. DataShield consists of three core components: (1) Compliance Vector Extraction, which captures the LLM's compliance behavior tendency; (2) a novel Compliance-Aware Score (CAS), which automatically identifies the optimal safety-critical layer; and (3) Safety-degrading Sample Filtering, which quantifies the projection shift of training data along the compliance direction. Extensive experimental evaluation on Llama3-8B, Llama3.1-8B, and Qwen2.5-7B using the Alpaca and Dolly benign datasets validates our method's effectiveness in identifying high-risk and low-risk data subsets. We also observe that open-ended question answering is more likely to trigger safety degradation, and corresponding responses tend to be longer. We hope this work can provide new insights into data-centric defense methods. The source code is available at: this https URL. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) Cite as: arXiv:2606.00160 [cs.CR]   (or arXiv:2606.00160v1 [cs.CR] for this version)   https://doi.org/10.48550/arXiv.2606.00160 Focus to learn more Submission history From: Junbo Zhang [view email] [v1] Fri, 29 May 2026 09:04:27 UTC (691 KB) Access Paper: HTML (experimental) view license Current browse context: cs.CR < prev   |   next > new | recent | 2026-06 Change to browse by: cs cs.AI cs.CL References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)