# Fight Fire with Fire: Defending Against Malicious RL Fine-Tuning via Reward Neutralization

arXiv: https://arxiv.org/abs/2505.04578 · HTML: https://mlsanity.com/paper/2505.04578/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Prompt injection and agent security, one of 12 directions in LLM safety and security.
- Builds on 28 of its 35 references that are in the map.
- First posted on arXiv 7 May 2025.

## Placed in

- LLM safety and security / [Prompt injection and agent security](https://mlsanity.com/safety/prompt-injection-and-agent-security/)

## Builds on (most cited first, 28 of 28)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Asynchronous Methods for Deep Reinforcement Learning](https://mlsanity.com/paper/1602.01783/) (2016)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [Safety Alignment Should Be Made More Than Just a Few Tokens Deep](https://mlsanity.com/paper/2406.05946/) (2024)
- [Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models](https://mlsanity.com/paper/2310.02949/) (2023)
- [Removing RLHF Protections in GPT-4 via Fine-Tuning](https://mlsanity.com/paper/2311.05553/) (2023)
- [Tamper-Resistant Safeguards for Open-Weight LLMs](https://mlsanity.com/paper/2408.00761/) (2024)
- [ODIN: Disentangled Reward Mitigates Hacking in RLHF](https://mlsanity.com/paper/2402.07319/) (2024)
- [Safety Layers in Aligned Large Language Models: The Key to LLM Security](https://mlsanity.com/paper/2408.17003/) (2024)
- [Vaccine: Perturbation-aware Alignment for Large Language Models against Harmful Fine-tuning Attack](https://mlsanity.com/paper/2402.01109/) (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)
- [DEPN: Detecting and Editing Privacy Neurons in Pretrained Language Models](https://mlsanity.com/paper/2310.20138/) (2023)
- [BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT](https://mlsanity.com/paper/2304.12298/) (2023)
- [Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis](https://mlsanity.com/paper/2406.10794/) (2024)
- [Unveiling the Implicit Toxicity in Large Language Models](https://mlsanity.com/paper/2311.17391/) (2023)
- [Enhancing LLM Safety via Constrained Direct Preference Optimization](https://mlsanity.com/paper/2403.02475/) (2024)
- [Immunization against harmful fine-tuning attacks](https://mlsanity.com/paper/2402.16382/) (2024)
- [The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking](https://mlsanity.com/paper/2501.19358/) (2025)
- [LLMStinger: Jailbreaking LLMs using RL fine-tuned LLMs](https://mlsanity.com/paper/2411.08862/) (2024)
- [Adversarial Agents: Black-Box Evasion Attacks with Reinforcement Learning](https://mlsanity.com/paper/2503.01734/) (2025)

## Built on by (most cited first, 3 of 3)

- [Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning](https://mlsanity.com/paper/2508.20697/) (2025)
- [Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink](https://mlsanity.com/paper/2602.05228/) (2026)
- [Learning to Stay Safe: Adaptive Regularization Against Safety Degradation during Fine-Tuning](https://mlsanity.com/paper/2602.17546/) (2026)

## Abstract (arXiv)

Reinforcement learning (RL) fine-tuning transforms large language models while creating a vulnerability we experimentally verify: Our experiment shows that malicious RL fine-tuning dismantles safety guardrails with remarkable efficiency, requiring only 50 steps and minimal adversarial prompts, with harmful escalating from 0-2 to 7-9. This attack vector particularly threatens open-source models with parameter-level access. Existing defenses targeting supervised fine-tuning prove ineffective against RL's dynamic feedback mechanisms. We introduce Reward Neutralization, the first defense framework specifically designed against RL fine-tuning attacks, establishing concise rejection patterns that render malicious reward signals ineffective. Our approach trains models to produce minimal-information rejections that attackers cannot exploit, systematically neutralizing attempts to optimize toward harmful outputs. Experiments validate that our approach maintains low harmful scores (no greater than 2) after 200 attack steps, while standard models rapidly deteriorate. This work provides the first constructive proof that robust defense against increasingly accessible RL attacks is achievable, addressing a critical security gap for open-weight models.