Safety-Potential Pruning for Enhancing Safety Prompts Against VLM Jailbreaking Without Retraining

• URL: http://arxiv.org/abs/2603.14219v1
• Date: Sun, 15 Mar 2026 04:45:08 GMT
• Title: Safety-Potential Pruning for Enhancing Safety Prompts Against VLM Jailbreaking Without Retraining
• Abstract summary: Safety prompts constitute an interpretable layer of defense against jailbreak attacks in vision-language models.<n>We introduce Safety-Potential Pruning, a one-shot pruning framework that amplifies safety-relevant activations.<n>Our method reduces attack success rates by up to 22% relative to prompting alone, all while maintaining strong benign performance.
• Score: 3.347328024911274
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Safety prompts constitute an interpretable layer of defense against jailbreak attacks in vision-language models (VLMs); however, their efficacy is constrained by the models' latent structural responsiveness. We observe that such prompts consistently engage a sparse set of parameters that remain largely quiescent during benign use. This finding motivates the Safety Subnetwork Hypothesis: VLMs embed structurally distinct pathways capable of enforcing safety, but these pathways remain dormant without explicit stimulation. To expose and amplify these pathways, we introduce Safety-Potential Pruning, a one-shot pruning framework that amplifies safety-relevant activations by removing weights that are less responsive to safety prompts without additional retraining. Across three representative VLM architectures and three jailbreak benchmarks, our method reduces attack success rates by up to 22% relative to prompting alone, all while maintaining strong benign performance. These findings frame pruning not only as a model compression technique, but as a structural intervention to emerge alignment-relevant subnets, offering a new path to robust jailbreak resistance.

Related papers

• Safety Paradox: How Enhanced Safety Awareness Leaves LLMs Vulnerable to Posterior Attack [19.201635180438313]
  Large language models (LLMs) are rigorously aligned to refuse harmful requests.<n>In this work, we reveal that this advanced safety awareness inadvertently introduces a fatal vulnerability.
  arXiv  Detail & Related papers  (2026-06-04T02:36:41Z)
• SafeThinker: Reasoning about Risk to Deepen Safety Beyond Shallow Alignment [43.86865924673546]
  We propose SafeThinker, an adaptive framework that allocates defensive resources via a lightweight gateway classifier.<n> Experiments show that SafeThinker significantly lowers attack success rates across diverse jailbreak strategies without compromising robustness.
  arXiv  Detail & Related papers  (2026-01-23T07:12:53Z)
• Attributing and Exploiting Safety Vectors through Global Optimization in Large Language Models [50.91504059485288]
  We propose a framework that identifies safety-critical attention heads through global optimization over all heads simultaneously.<n>We develop a novel inference-time white-box jailbreak method that exploits the identified safety vectors through activation repatching.
  arXiv  Detail & Related papers  (2026-01-22T09:32:43Z)
• Understanding and Preserving Safety in Fine-Tuned LLMs [20.821783178639063]
  Fine-tuning can substantially degrade safety alignment, even when the fine-tuning data is harmless.<n>We propose safety-preserving fine-tuning (SPF), a lightweight approach that explicitly removes gradient components conflicting with the low-rank safety subspace.<n> SPF consistently maintains downstream task performance and recovers nearly all pre-trained safety alignment, even under adversarial fine-tuning scenarios.
  arXiv  Detail & Related papers  (2026-01-15T07:33:13Z)
• SafeR-CLIP: Mitigating NSFW Content in Vision-Language Models While Preserving Pre-Trained Knowledge [51.634837361795434]
  SaFeR-CLIP reconciles safety and performance, recovering up to 8.0% in zero-shot accuracy over prior methods.<n>We also contribute NSFW-Caps, a new benchmark of 1,000 highly-aligned pairs for testing safety under distributional shift.
  arXiv  Detail & Related papers  (2025-11-20T19:00:15Z)
• EASE: Practical and Efficient Safety Alignment for Small Language Models [4.839980912290382]
  Small language models (SLMs) are increasingly deployed on edge devices, making their safety alignment crucial yet challenging.<n>We propose EASE, a novel framework that enables practical and Efficient safety alignment for Small languagE models.
  arXiv  Detail & Related papers  (2025-11-09T19:46:54Z)
• Toward Safer Diffusion Language Models: Discovery and Mitigation of Priming Vulnerability [5.650647159993238]
  Diffusion language models (DLMs) generate tokens in parallel through iterative denoising.<n>In this paper, we reveal that DLMs have a critical vulnerability stemming from their iterative denoising process.<n>We propose a novel safety alignment method tailored to DLMs that trains models to generate safe responses from contaminated intermediate states.
  arXiv  Detail & Related papers  (2025-10-01T06:35:23Z)
• Beyond Surface Alignment: Rebuilding LLMs Safety Mechanism via Probabilistically Ablating Refusal Direction [21.03567306455414]
  Jailbreak attacks pose persistent threats to large language models (LLMs)<n>We introduce DeepRefusal, a robust safety alignment framework that overcomes these issues.<n>Our method reduces attack success rates by approximately 95%, while maintaining model capabilities with minimal performance degradation.
  arXiv  Detail & Related papers  (2025-09-18T17:54:31Z)
• Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets [64.96967819446553]
  This paper investigates the degradation of safety guardrails through the lens of representation similarity between upstream alignment datasets and downstream fine-tuning tasks.<n>High similarity between these datasets significantly weakens safety guardrails, making models more susceptible to jailbreaks.<n>Low similarity between these two types of datasets yields substantially more robust models and thus reduces harmfulness score by up to 10.33%.
  arXiv  Detail & Related papers  (2025-06-05T17:59:55Z)
• Reasoning-to-Defend: Safety-Aware Reasoning Can Defend Large Language Models from Jailbreaking [54.10710423370126]
  We propose Reasoning-to-Defend (R2D), a training paradigm that integrates a safety-aware reasoning mechanism into Large Language Models' generation process.<n>CPO enhances the model's perception of the safety status of given dialogues.<n>Experiments demonstrate that R2D effectively mitigates various attacks and improves overall safety, while maintaining the original performances.
  arXiv  Detail & Related papers  (2025-02-18T15:48:46Z)
• The VLLM Safety Paradox: Dual Ease in Jailbreak Attack and Defense [56.32083100401117]
  The vulnerability of Vision Large Language Models (VLLMs) to jailbreak attacks appears as no surprise.<n>Recent defense mechanisms against these attacks have reached near-saturation performance on benchmark evaluations.
  arXiv  Detail & Related papers  (2024-11-13T07:57:19Z)
• Defensive Prompt Patch: A Robust and Interpretable Defense of LLMs against Jailbreak Attacks [59.46556573924901]
  This paper introduces Defensive Prompt Patch (DPP), a novel prompt-based defense mechanism for large language models (LLMs)<n>Unlike previous approaches, DPP is designed to achieve a minimal Attack Success Rate (ASR) while preserving the high utility of LLMs.<n> Empirical results conducted on LLAMA-2-7B-Chat and Mistral-7B-Instruct-v0.2 models demonstrate the robustness and adaptability of DPP.
  arXiv  Detail & Related papers  (2024-05-30T14:40:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.