# Retention Score: Quantifying Jailbreak Risks for Vision Language Models

arXiv: https://arxiv.org/abs/2412.17544 · HTML: https://mlsanity.com/paper/2412.17544/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 15 of its 44 references that are in the map.
- First posted on arXiv 23 Dec 2024.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM jailbreaks and adversarial robustness](https://mlsanity.com/mllm/mllm-jailbreaks-and-adversarial-robustness/)
- LLM safety and security / [Vision-language and text-to-image safety](https://mlsanity.com/safety/vision-language-and-text-to-image-safety/)

## Builds on (most cited first, 15 of 15)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [EVA-CLIP: Improved Training Techniques for CLIP at Scale](https://mlsanity.com/paper/2303.15389/) (2023)
- [EVA: Exploring the Limits of Masked Visual Representation Learning at Scale](https://mlsanity.com/paper/2211.07636/) (2022)
- [On Evaluating Adversarial Robustness](https://mlsanity.com/paper/1902.06705/) (2019)
- [Visual Adversarial Examples Jailbreak Aligned Large Language Models](https://mlsanity.com/paper/2306.13213/) (2023)
- [Are aligned neural networks adversarially aligned?](https://mlsanity.com/paper/2306.15447/) (2023)
- [Evaluating the Robustness of Neural Networks: An Extreme Value Theory Approach](https://mlsanity.com/paper/1801.10578/) (2018)

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

- [A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment](https://mlsanity.com/paper/2504.15585/) (2025)
- [The Safeguard Worked. Is the LLM System Safer?](https://mlsanity.com/paper/2609.00519/) (2026)

## Abstract (arXiv)

The emergence of Vision-Language Models (VLMs) is a significant advancement in integrating computer vision with Large Language Models (LLMs) to enhance multi-modal machine learning capabilities. However, this progress has also made VLMs vulnerable to sophisticated adversarial attacks, raising concerns about their reliability. The objective of this paper is 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 Retention Score. Retention Score is a multi-modal evaluation metric that includes Retention-I and Retention-T scores for quantifying jailbreak risks in visual and textual components of VLMs. Our process involves generating synthetic image-text pairs using a conditional diffusion model. These pairs are then predicted for toxicity score by a VLM alongside a toxicity judgment classifier. By calculating the margin in toxicity scores, we can quantify the robustness of the VLM in an attack-agnostic manner. Our work has four main contributions. First, we prove that Retention Score can serve as a certified robustness metric. Second, we demonstrate that most VLMs with visual components are less robust against jailbreak attacks than the corresponding plain VLMs. Additionally, we evaluate black-box VLM APIs and find that the security settings in Google Gemini significantly affect the score and robustness. Moreover, the robustness of GPT4V is similar to the medium settings of Gemini. Finally, our approach offers a time-efficient alternative to existing adversarial attack methods and provides consistent model robustness rankings when evaluated on VLMs including MiniGPT-4, InstructBLIP, and LLaVA.