Safety-Preserving Fine-Tuning of Language Models

Ask about this topic

Latest papers

CardsList

57 papers25 papers in the last four weeks, up 733% on the four weeks before. 0.2% of all new papers.

  1. Is Persona Generalization Predictable?

    Nov 9, 2026Olina Mukherjee, Arnav Goel, Aditi Raghunathan +1LLM Fine-TuningSafety-Preserving Fine-Tuning of Language Models

    Carnegie Mellon University

  2. FireGate: Bounding Behaviour Change Before Training in Continual RL Fine-Tuning

    Nov 9, 2026Md Anwar Hossen, Ali JannesariFine-Tuning Language ModelsSafety-Preserving Fine-Tuning of Language Models

    Department of Computer Science, Iowa State University, Ames, IA, USA

  3. TLRS-DPO: Token-Level Risk Scoring for Better Direct Preference Optimization

    Nov 8, 2026Suxin Ji, Hungtao WanDirect Preference OptimizationSafety-Preserving Fine-Tuning of Language Models

    University of Pennsylvania · Independent Researcher

  4. When Safety Becomes An Outlier: Understanding the Retention of LLM Safety Behaviors

    Nov 7, 2026Binchi Zhang, Hadi Abdullah, Yiwei Cai +1Safety-Preserving Fine-Tuning of Language ModelsLLM Safety Alignment

    University of Virginia · Visa Research

  5. Restoring General and Improving Domain-Specific Safety in Fine-Tuned LLMs Without Retraining

    Nov 6, 2026Vishnu Kabir Chhabra, Ding Zhu, Zhihui Zhu +2LLM Safety AlignmentSafety-Preserving Fine-Tuning of Language Models

    The Ohio State University, Columbus, OH

  6. SAVeR2^2: Reasoning-based Safety Alignment for Large Reasoning Models via Verifiable Rewards

    Nov 5, 2026Yichen Sun, Qirui Liu, Yan Wang +5LLM Safety AlignmentSafety-Preserving Fine-Tuning of Language Models

    The State Key Laboratory of Blockchain and Data Security, Zhejiang University · Alibaba Group · Chongqing Ant Consumer Finance Co., Ltd. +1

  7. Addition by Subtraction: Toward Mitigating Fragile Safety Alignment in Large Language Models

    Nov 4, 2026Hamid Dashtbani, Alexandra DmitrienkoLLM Safety AlignmentSafety-Preserving Fine-Tuning of Language Models

    Department of Computer Science, University of Duisburg-Essen

  8. Suppressing Misaligned Behaviors in Language Models, Durably

    Nov 3, 2026Luxi He, Pengcheng Jiang, Jifan Zhang +2LLM AlignmentLLM Unlearning

    Princeton University · UIUC · Thinking Machines Lab

  9. Neologism Inoculation Prompting

    Nov 3, 2026David Demitri Africa, Cameron HolmesSafety-Preserving Fine-Tuning of Language ModelsPrompt Injection Defense

    Resolution · UK AI Security Institute

  10. SecPO: Principled Adversarial Training for Prompt Injection Security

    Oct 30, 2026Sizhe Chen, Yibo Peng, Jaewon Chang +2Prompt Injection DefensePrompt Injection Attacks on AI Agents

    UC Berkeley · Anthropic · *work done at Google DeepMind

  11. ShallowDefend: Harnessing Shallow Safety Alignment for Multi-Turn Defender

    Oct 29, 2026Qiyao Ma, Xiaodong Liu, Weiwei Yang +9LLM Safety AlignmentJailbreak Defense

    University of California, Davis · Microsoft Research

  12. Safin-1.5: Safety-Native Architecture and Verifiable Reasoning for Recursive Safety Improvement

    Oct 29, 2026Jingnan Zheng, Cheng Jin, Ming Zhang +28LLM SafetyLanguage Model Self-Improvement

    Shanghai AI Laboratory · Hugging Face

  13. Revisiting Over-refusal in Safety-aligned Large Language Models: A Representation and Optimization Perspective

    Oct 29, 2026Chao Chen, Pengyang Shao, Wei Qin +3Gradient OrthogonalizationOver-Refusal

    Hefei University of Technology Hefei, China · National University of Singapore Singapore, Singapore

  14. Adversarial Self-Distillation

    Oct 28, 2026Mohammad Sadegh Akhondzadeh, Soroush H. Zargarbashi, Aleksandar BojchevskiAdversarial Robustness of Language ModelsSafety-Preserving Fine-Tuning of Language Models

    University of Cologne · CISPA Helmholtz Center for Information Security

  15. Post-Training Large Language Models with LLMON

    Oct 28, 2026Basel Shbita, Michael Hind, Chad DeLuca +8LLM Post-TrainingPrompt Injection Defense

    IBM Research, San Jose, CA, USA · IBM Research, Yorktown Heights, NY, USA · IBM Research, Cambridge, MA, USA +1

  16. Preserving Safety During Sequential Fine-Tuning: A Factorial and Geometric Analysis of Replay

    Oct 27, 2026George Ruellan, Diana Benavides-PradoSafety-Preserving Fine-Tuning of Language ModelsSafety-Utility Trade-Off in Language Models

    School of Electronic Engineering and Computer Science, Queen Mary University of London, London, United Kingdom

  17. Follow the Gradient: Tracing Safety Degradation Back to Training Data

    Oct 27, 2026Saket Reddy, Amro Abbas, Anshuman Suri +3Safety-Preserving Fine-Tuning of Language ModelsGradient-Based Attribution

    DatologyAI · University of Illinois Urbana-Champaign (UIUC)

  18. Amplify, Locate, Estimate (ALE) on Rare-Failure Rates of Fine-Tuned LLMs

    Oct 26, 2026Patrick Reichherzer, Jonathan Evertz, Thorsten HolzRare-Event EstimationLanguage Model Safety Evaluation

    Max Planck Institute for Security and Privacy (MPI-SP) · CISPA Helmholtz Center for Information Security

  19. SLoSa: Small, Low-Power, and Safe Foundation Models for Robotic Task Planning

    Oct 20, 2026Weizhe Xu, Jialiang Fan, Essien Taylor +3Robot Task PlanningLarge Language Model-Based Robot Planning

    University of Notre Dame · Northwestern University

  20. Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models

    Oct 20, 2026Niccolò RonchettiLLM UnlearningMachine Unlearning

  21. Safety Mid-Training: Internalizing LLM Safety as a Foundational Capability

    Oct 19, 2026Zhengyue Zhao, Liwei Jiang, Yejin Choi +1Safety-Utility Trade-Off in Language ModelsLLM Safety Alignment

    Johns Hopkins University · University of Washington · Stanford University

  22. AfriGuard: Safety and Robust Refusal For African Languages

    Oct 19, 2026Israel Abebe Azime, Jesujoba O. Alabi, Iffat Maab +20LLM SafetyLanguage Model Safety Evaluation

    Saarland University, Saarland Informatic Campus, Germany · Masakhane NLP · National Institute of Informatics, Japan +13

  23. Rethinking LLM Fine-Tuning via Weight Space Reparameterization: Preserving Safety during Downstream Adaptation

    Oct 19, 2026Min-Seong Kim, Jongbok Won, Jeesup Park +2Safety-Preserving Fine-Tuning of Language ModelsLLM Fine-Tuning

    Yonsei University

  24. When Safety Fails Collectively: Robust Alignment Against Harmful Fine-Tuning via Joint Weight Perturbations

    Oct 19, 2026Zhiyuan Wang, Yuezhi Zhou, Zhu Zhang +1Safety-Preserving Fine-Tuning of Language ModelsAdversarial Robustness of Language Models

    Dept. of Comp. Sci. & Tech., BNRist, Tsinghua University · Zhongguancun Laboratory

  25. Safety Kernel: Mechanistic Alignment for Cross-Lingual Defense in LLMs

    Oct 17, 2026Chengyu Xie, Fei Shen, Shuyi Miao +5LLM SafetyLanguage Model Robustness

    School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China · Nanjing University of Science and Technology, Nanjing, China · School of Artificial Intelligence, Beihang University, Beijing, China +3

  26. Defending Against Dormant Poisoning Attacks Across Language and Multimodal Agents

    Oct 17, 2026Wonjun Lee, Vaidehi Patil, Zaid Khan +4AI Agent SafetyBackdoor Defense in LLMs

    UNC Chapel Hill · Kyung Hee University

  27. Rethinking GRPO for Large Reasoning Model Safety Alignment

    Oct 16, 2026Jiawei Chen, Yi Zhang, Jiawei Du +9Safety-Preserving Fine-Tuning of Language ModelsLLM Safety Alignment

    East China Normal University · Baidu Inc. · Zhongguancun Academy +5

  28. TRAP: Understanding and Mitigating Privacy Memorization in Language Models

    Sep 26, 2026Muhammed Ustaomeroglu, Ziyue Xu, Hanshen Xiao +3Safety-Preserving Fine-Tuning of Language ModelsMemorization in Language Models

    Carnegie Mellon University · NVIDIA · Purdue University

  29. Beyond Average Safety: Chance-Constrained LLM Fine-tuning

    Sep 24, 2026Taha Entesari, Mahyar FazlyabStochastic OptimizationSafety-Preserving Fine-Tuning of Language Models

    Department of Electrical and Computer Engineering Johns Hopkins University

  30. Inoculation Midtraining with Learned Neologisms

    Sep 14, 2026Kyle O'Brien, Edward James Young, Puria Radmard +4LLM AlignmentLLM Safety

    Geodesic Research · OpenAI · UK AI Security Institute