Zero-Shot Defense Against Toxic Images via Inherent Multimodal Alignment in LVLMs

• URL: http://arxiv.org/abs/2503.00037v1
• Date: Tue, 25 Feb 2025 06:51:16 GMT
• Title: Zero-Shot Defense Against Toxic Images via Inherent Multimodal Alignment in LVLMs
• Authors: Wei Zhao, Zhe Li, Yige Li, Jun Sun,
• Abstract summary: We introduce SafeCLIP, a lightweight method that leverages LVLMs inherent multimodal alignment for zero-shot toxic image detection.<n>Experiments show that SafeCLIP achieves a 66.9% defense success rate with only 3.2% false positive rate and 7.2% overhead.<n>Our work demonstrates that leveraging inherent multimodal alignment can yield efficient, low-cost LVLM safety.
• Score: 10.463762448166714
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Vision-Language Models (LVLMs) have made significant strides in multimodal comprehension, thanks to extensive pre-training and fine-tuning on large-scale visual datasets. However, despite their robust textual safety mechanisms, they remain vulnerable to harmful visual inputs. Existing safeguards-typically relying on pre-filtering or fine-tuning-incur high costs and diminish overall utility. To address this critical vulnerability, we introduce SafeCLIP, a lightweight method that leverages LVLMs inherent multimodal alignment for zero-shot toxic image detection. By projecting CLIPs discarded CLS token into its text space and matching it with toxic descriptors, SafeCLIP detects harmful content without any architectural changes-adding minimal latency and enabling dynamic safety corrections during inference and fine-tuning.Experiments show that SafeCLIP achieves a 66.9% defense success rate with only 3.2% false positive rate and 7.2% overhead. In contrast, state-of-the-art methods achieve 52.9% success but have a 10.7% false positive rate and 210% overhead. Our work demonstrates that leveraging inherent multimodal alignment can yield efficient, low-cost LVLM safety. Code is available at anonymous.4open.science/r/safeclip-2C01.

Related papers

• Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models? [83.53005932513155]
  Multi-modal large language models (MLLMs) have made significant progress, yet their safety alignment remains limited. We propose finetuning MLLMs on a small set of benign instruct-following data with responses replaced by simple, clear rejection sentences.
  arXiv  Detail & Related papers  (2025-04-14T09:03:51Z)
• Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-tuning [23.71517734919702]
  Vision-language models (VLMs) have made remarkable strides in generative modeling with multimodal inputs. Current alignment strategies rely on supervised safety fine-tuning with curated datasets. We show that supervised fine-tuning inadvertently reinforces spurious correlations between superficial textual patterns and safety responses.
  arXiv  Detail & Related papers  (2025-03-14T19:52:08Z)
• HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden States [17.601328965546617]
  We investigate whether LVLMs inherently encode safety-relevant signals within their internal activations during inference.<n>Our findings reveal that LVLMs exhibit distinct activation patterns when processing unsafe prompts.<n>We introduce HiddenDetect, a novel tuning-free framework that harnesses internal model activations to enhance safety.
  arXiv  Detail & Related papers  (2025-02-20T17:14:34Z)
• Internal Activation as the Polar Star for Steering Unsafe LLM Behavior [50.463399903987245]
  We introduce SafeSwitch, a framework that dynamically regulates unsafe outputs by monitoring and utilizing the model's internal states. Our empirical results show that SafeSwitch reduces harmful outputs by over 80% on safety benchmarks while maintaining strong utility.
  arXiv  Detail & Related papers  (2025-02-03T04:23:33Z)
• Defending Multimodal Backdoored Models by Repulsive Visual Prompt Tuning [19.638259197558625]
  Multimodal contrastive learning models (e.g., CLIP) can learn high-quality representations from large-scale image-text datasets. They exhibit significant vulnerabilities to backdoor attacks, raising serious safety concerns. We propose Repulsive Visual Prompt Tuning (RVPT) as a novel defense approach.
  arXiv  Detail & Related papers  (2024-12-29T08:09:20Z)
• Retention Score: Quantifying Jailbreak Risks for Vision Language Models [60.48306899271866]
  Vision-Language Models (VLMs) are integrated with Large Language Models (LLMs) to enhance multi-modal machine learning capabilities. This paper aims to assess the resilience of VLMs against jailbreak attacks that can compromise model safety compliance and result in harmful outputs. To evaluate a VLM's ability to maintain its robustness against adversarial input perturbations, we propose a novel metric called the textbfRetention Score.
  arXiv  Detail & Related papers  (2024-12-23T13:05:51Z)
• Exploring Visual Vulnerabilities via Multi-Loss Adversarial Search for Jailbreaking Vision-Language Models [92.79804303337522]
  Vision-Language Models (VLMs) may still be vulnerable to safety alignment issues.<n>We introduce MLAI, a novel jailbreak framework that leverages scenario-aware image generation for semantic alignment.<n>Extensive experiments demonstrate MLAI's significant impact, achieving attack success rates of 77.75% on MiniGPT-4 and 82.80% on LLaVA-2.
  arXiv  Detail & Related papers  (2024-11-27T02:40:29Z)
• CoCA: Regaining Safety-awareness of Multimodal Large Language Models with Constitutional Calibration [90.36429361299807]
  multimodal large language models (MLLMs) have demonstrated remarkable success in engaging in conversations involving visual inputs. The integration of visual modality has introduced a unique vulnerability: the MLLM becomes susceptible to malicious visual inputs. We introduce a technique termed CoCA, which amplifies the safety-awareness of the MLLM by calibrating its output distribution.
  arXiv  Detail & Related papers  (2024-09-17T17:14:41Z)
• Prompt Leakage effect and defense strategies for multi-turn LLM interactions [95.33778028192593]
  Leakage of system prompts may compromise intellectual property and act as adversarial reconnaissance for an attacker. We design a unique threat model which leverages the LLM sycophancy effect and elevates the average attack success rate (ASR) from 17.7% to 86.2% in a multi-turn setting. We measure the mitigation effect of 7 black-box defense strategies, along with finetuning an open-source model to defend against leakage attempts.
  arXiv  Detail & Related papers  (2024-04-24T23:39:58Z)
• Uncovering Safety Risks of Large Language Models through Concept Activation Vector [13.804245297233454]
  We introduce a Safety Concept Activation Vector (SCAV) framework to guide attacks on large language models (LLMs)<n>We then develop an SCAV-guided attack method that can generate both attack prompts and embedding-level attacks.<n>Our attack method significantly improves the attack success rate and response quality while requiring less training data.
  arXiv  Detail & Related papers  (2024-04-18T09:46:25Z)
• Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation [98.02846901473697]
  We propose ECSO (Eyes Closed, Safety On), a training-free protecting approach that exploits the inherent safety awareness of MLLMs. ECSO generates safer responses via adaptively transforming unsafe images into texts to activate the intrinsic safety mechanism of pre-aligned LLMs.
  arXiv  Detail & Related papers  (2024-03-14T17:03:04Z)
• FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts [14.33139608409507]
  We propose FigStep, a straightforward yet effective black-box jailbreak algorithm against LVLMs.<n>FigStep converts the prohibited content into images through typography to bypass the safety alignment.<n>Our work reveals that current LVLMs are vulnerable to jailbreak attacks, which highlights the necessity of novel cross-modality safety alignment techniques.
  arXiv  Detail & Related papers  (2023-11-09T18:59:11Z)

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.