# Multi-Level Safety Continual Projection for Fine-Tuned Large Language Models without Retraining

arXiv: https://arxiv.org/abs/2508.09190 · HTML: https://mlsanity.com/paper/2508.09190/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security.
- Builds on 30 of its 39 references that are in the map.
- First posted on arXiv 8 Aug 2025.

## Placed in

- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

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

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Jailbreaking Black Box Large Language Models in Twenty Queries](https://mlsanity.com/paper/2310.08419/) (2023)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations](https://mlsanity.com/paper/2310.06387/) (2023)
- [Detecting Language Model Attacks with Perplexity](https://mlsanity.com/paper/2308.14132/) (2023)
- [Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions](https://mlsanity.com/paper/2309.07875/) (2023)
- [LLM Self Defense: By Self Examination, LLMs Know They Are Being Tricked](https://mlsanity.com/paper/2308.07308/) (2023)
- [Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications](https://mlsanity.com/paper/2402.05162/) (2024)
- [Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization](https://mlsanity.com/paper/2311.09096/) (2023)
- [Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models](https://mlsanity.com/paper/2402.02207/) (2024)
- [Tamper-Resistant Safeguards for Open-Weight LLMs](https://mlsanity.com/paper/2408.00761/) (2024)
- [Safety Layers in Aligned Large Language Models: The Key to LLM Security](https://mlsanity.com/paper/2408.17003/) (2024)
- [Safe LoRA: the Silver Lining of Reducing Safety Risks when Fine-tuning Large Language Models](https://mlsanity.com/paper/2405.16833/) (2024)
- [Vaccine: Perturbation-aware Alignment for Large Language Models against Harmful Fine-tuning Attack](https://mlsanity.com/paper/2402.01109/) (2024)
- [Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack](https://mlsanity.com/paper/2405.18641/) (2024)
- [Defending Against Unforeseen Failure Modes with Latent Adversarial Training](https://mlsanity.com/paper/2403.05030/) (2024)
- [Antidote: Post-fine-tuning Safety Alignment for Large Language Models against Harmful Fine-tuning](https://mlsanity.com/paper/2408.09600/) (2024)
- [Privacy Issues in Large Language Models: A Survey](https://mlsanity.com/paper/2312.06717/) (2023)
- [NLSR: Neuron-Level Safety Realignment of Large Language Models Against Harmful Fine-Tuning](https://mlsanity.com/paper/2412.12497/) (2024)
- [SafeMERGE: Preserving Safety Alignment in Fine-Tuned Large Language Models via Selective Layer-Wise Model Merging](https://mlsanity.com/paper/2503.17239/) (2025)
- [Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models](https://mlsanity.com/paper/2410.02298/) (2024)
- [Locking Down the Finetuned LLMs Safety](https://mlsanity.com/paper/2410.10343/) (2024)
- [Safe Pruning LoRA: Robust Distance-Guided Pruning for Safety Alignment in Adaptation of LLMs](https://mlsanity.com/paper/2506.18931/) (2025)
- [What do Large Language Models Say About Animals? Investigating Risks of Animal Harm in Generated Text](https://mlsanity.com/paper/2503.04804/) (2025)
- [PiCo: Jailbreaking Multimodal Large Language Models via Pictorial Code Contextualization](https://mlsanity.com/paper/2504.01444/) (2025)

## Built on by (most cited first, 2 of 2)

- [Targeted Interpretable Safety Neuron Enhancement for Multilingual Vision-Language Large Models](https://mlsanity.com/paper/2604.08881/) (2026)
- [One Anchor for All: Unified Multilingual and Multimodal Safety Alignment for LVLMs](https://mlsanity.com/paper/2607.27917/) (2026)

## Abstract (arXiv)

While fine-tuning services drive the rapid expansion of task capabilities in large language models (LLMs), they are often accompanied by the degradation and reorganization of safety-aligned representations, making models more prone to deviating from human preferences and exposing them to emerging jailbreak risks. Existing post-fine-tuning defense methods predominantly rely on single-scale safety correction mechanisms, which struggle to achieve a robust balance among safety, model utility, and continual adaptability. We propose Multi-Level Safety Continual Projection (MSCP), a training-free post-fine-tuning safety enhancement method that implicitly aligns global and localized safety activations through coordinated multi-level representations to isolate sparse neuron clusters governing safety-sensitive behaviors. It then applies composable safety-direction projections without retraining, effectively suppressing harmful outputs under minimal parameter perturbations while preserving task performance and improving alignment with human preferences. Extensive experiments across multiple fine-tuned LLM models demonstrate that our method significantly reduce harmfulness scores and attack success rates with minimal parameter modifications, while preserving the model's utility. Furthermore, we introduce a task-specific, multi-dimensional heterogeneous safety activation clustering mechanism that enables continual defense and generalization capability against unforeseen emerging safety concerns.