# Immunization against harmful fine-tuning attacks

arXiv: https://arxiv.org/abs/2402.16382 · HTML: https://mlsanity.com/paper/2402.16382/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- Builds on 49 of its 65 references that are in the map.
- First posted on arXiv 26 Feb 2024.

## Builds on (most cited first, 30 of 49)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks](https://mlsanity.com/paper/1703.03400/) (2017)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples](https://mlsanity.com/paper/1802.00420/) (2018)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations](https://mlsanity.com/paper/2312.06674/) (2023)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [On First-Order Meta-Learning Algorithms](https://mlsanity.com/paper/1803.02999/) (2018)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training](https://mlsanity.com/paper/2401.05566/) (2024)
- [DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models](https://mlsanity.com/paper/2306.11698/) (2023)
- [Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models](https://mlsanity.com/paper/2310.02949/) (2023)
- [Poisoning Language Models During Instruction Tuning](https://mlsanity.com/paper/2305.00944/) (2023)
- [LEACE: Perfect linear concept erasure in closed form](https://mlsanity.com/paper/2306.03819/) (2023)
- [The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics](https://mlsanity.com/paper/2102.01672/) (2021)
- [Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications](https://mlsanity.com/paper/2402.05162/) (2024)
- [LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B](https://mlsanity.com/paper/2310.20624/) (2023)
- [A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity](https://mlsanity.com/paper/2401.01967/) (2024)
- [Removing RLHF Protections in GPT-4 via Fine-Tuning](https://mlsanity.com/paper/2311.05553/) (2023)
- [Towards Deep Neural Network Architectures Robust to Adversarial Examples](https://mlsanity.com/paper/1412.5068/) (2014)
- [Attacks, Defenses and Evaluations for LLM Conversation Safety: A Survey](https://mlsanity.com/paper/2402.09283/) (2024)
- [Universal Jailbreak Backdoors from Poisoned Human Feedback](https://mlsanity.com/paper/2311.14455/) (2023)
- [On the Exploitability of Instruction Tuning](https://mlsanity.com/paper/2306.17194/) (2023)
- [Mechanistically analyzing the effects of fine-tuning on procedurally defined tasks](https://mlsanity.com/paper/2311.12786/) (2023)

## Built on by (most cited first, 30 of 35)

- [Tamper-Resistant Safeguards for Open-Weight LLMs](https://mlsanity.com/paper/2408.00761/) (2024)
- [Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates](https://mlsanity.com/paper/2402.18540/) (2024)
- [Representation Noising: A Defence Mechanism Against Harmful Finetuning](https://mlsanity.com/paper/2405.14577/) (2024)
- [Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack](https://mlsanity.com/paper/2405.18641/) (2024)
- [Booster: Tackling Harmful Fine-tuning for Large Language Models via Attenuating Harmful Perturbation](https://mlsanity.com/paper/2409.01586/) (2024)
- [Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Survey](https://mlsanity.com/paper/2409.18169/) (2024)
- [Antidote: Post-fine-tuning Safety Alignment for Large Language Models against Harmful Fine-tuning](https://mlsanity.com/paper/2408.09600/) (2024)
- [Operationalizing a Threat Model for Red-Teaming Large Language Models (LLMs)](https://mlsanity.com/paper/2407.14937/) (2024)
- [Targeted Vaccine: Safety Alignment for Large Language Models against Harmful Fine-Tuning via Layer-wise Perturbation](https://mlsanity.com/paper/2410.09760/) (2024)
- [No Two Devils Alike: Unveiling Distinct Mechanisms of Fine-tuning Attacks](https://mlsanity.com/paper/2405.16229/) (2024)
- [Panacea: Mitigating Harmful Fine-tuning for Large Language Models via Post-fine-tuning Perturbation](https://mlsanity.com/paper/2501.18100/) (2025)
- [Probe before You Talk: Towards Black-box Defense against Backdoor Unalignment for Large Language Models](https://mlsanity.com/paper/2506.16447/) (2025)
- [Vulnerability-Aware Alignment: Mitigating Uneven Forgetting in Harmful Fine-Tuning](https://mlsanity.com/paper/2506.03850/) (2025)
- [An Embarrassingly Simple Defense Against LLM Abliteration Attacks](https://mlsanity.com/paper/2505.19056/) (2025)
- [Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Scheduler](https://mlsanity.com/paper/2510.27172/) (2025)
- [CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning](https://mlsanity.com/paper/2505.16559/) (2025)
- [Gradient Surgery for Safe LLM Fine-Tuning](https://mlsanity.com/paper/2508.07172/) (2025)
- [Fundamental Limitations in Pointwise Defences of LLM Finetuning APIs](https://mlsanity.com/paper/2502.14828/) (2025)
- [Towards Resilient Safety-driven Unlearning for Diffusion Models against Downstream Fine-tuning](https://mlsanity.com/paper/2507.16302/) (2025)
- [MetaDefense: Defending Finetuning-based Jailbreak Attack Before and During Generation](https://mlsanity.com/paper/2510.07835/) (2025)
- [Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning](https://mlsanity.com/paper/2510.10085/) (2025)
- [Evaluating Defences against Unsafe Feedback in RLHF](https://mlsanity.com/paper/2409.12914/) (2024)
- [Elephant in the Room: Unveiling the Impact of Reward Model Quality in Alignment](https://mlsanity.com/paper/2409.19024/) (2024)
- [Fight Fire with Fire: Defending Against Malicious RL Fine-Tuning via Reward Neutralization](https://mlsanity.com/paper/2505.04578/) (2025)
- [Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems](https://mlsanity.com/paper/2410.23472/) (2024)
- [Consiglieres in the Shadow: Understanding the Use of Uncensored Large Language Models in Cybercrimes](https://mlsanity.com/paper/2508.12622/) (2025)
- [Layered LLM Defenses as an Ensemble: Access Tiers, Inference Cost, and the Measured Failure Correlation Between Defense Layers](https://mlsanity.com/paper/2608.28327/) (2026)
- [SafeTuneBed: A Toolkit for Benchmarking LLM Safety Alignment in Fine-Tuning](https://mlsanity.com/paper/2506.00676/) (2025)
- [Making Models Unmergeable via Scaling-Sensitive Loss Landscape](https://mlsanity.com/paper/2601.21898/) (2026)
- [AI Security in the Foundation Model Era: A Comprehensive Survey from a Unified Perspective](https://mlsanity.com/paper/2603.24857/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) are often trained with safety guards intended to prevent harmful text generation. However, such safety training can be removed by fine-tuning the LLM on harmful datasets. While this emerging threat (harmful fine-tuning attacks) has been characterized by previous work, there is little understanding of how we should proceed in constructing and validating defenses against these attacks especially in the case where defenders would not have control of the fine-tuning process. We introduce a formal framework based on the training budget of an attacker which we call "Immunization" conditions. Using a formal characterisation of the harmful fine-tuning problem, we provide a thorough description of what a successful defense must comprise of and establish a set of guidelines on how rigorous defense research that gives us confidence should proceed.