# Image Hijacks: Adversarial Images can Control Generative Models at Runtime

arXiv: https://arxiv.org/abs/2309.00236 · HTML: https://mlsanity.com/paper/2309.00236/

## Key facts

- Cited by 172 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 53 references that are in the map.
- First posted on arXiv 1 Sep 2023.

## Builds on (most cited first, 30 of 34)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Explaining and Harnessing Adversarial Examples](https://mlsanity.com/paper/1412.6572/) (2014)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [Intriguing properties of neural networks](https://mlsanity.com/paper/1312.6199/) (2013)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [Jailbroken: How Does LLM Safety Training Fail?](https://mlsanity.com/paper/2307.02483/) (2023)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [Certified Adversarial Robustness via Randomized Smoothing](https://mlsanity.com/paper/1902.02918/) (2019)
- [Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples](https://mlsanity.com/paper/1802.00420/) (2018)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images](https://mlsanity.com/paper/1412.1897/) (2014)
- [RobustBench: a standardized adversarial robustness benchmark](https://mlsanity.com/paper/2010.09670/) (2020)
- [Prompt Injection attack against LLM-integrated Applications](https://mlsanity.com/paper/2306.05499/) (2023)
- [Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learning](https://mlsanity.com/paper/2203.02053/) (2022)
- [Countering Adversarial Images using Input Transformations](https://mlsanity.com/paper/1711.00117/) (2017)
- [Adversarial Patch](https://mlsanity.com/paper/1712.09665/) (2017)
- [BERT-ATTACK: Adversarial Attack Against BERT Using BERT](https://mlsanity.com/paper/2004.09984/) (2020)
- [Augmented Language Models: a Survey](https://mlsanity.com/paper/2302.07842/) (2023)
- [Are aligned neural networks adversarially aligned?](https://mlsanity.com/paper/2306.15447/) (2023)
- [On Evaluating Adversarial Robustness of Large Vision-Language Models](https://mlsanity.com/paper/2305.16934/) (2023)
- [Uncovering the Limits of Adversarial Training against Norm-Bounded Adversarial Examples](https://mlsanity.com/paper/2010.03593/) (2020)
- [Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models](https://mlsanity.com/paper/2307.14539/) (2023)
- [Better Diffusion Models Further Improve Adversarial Training](https://mlsanity.com/paper/2302.04638/) (2023)
- [Automatically Auditing Large Language Models via Discrete Optimization](https://mlsanity.com/paper/2303.04381/) (2023)
- [On the Adversarial Robustness of Multi-Modal Foundation Models](https://mlsanity.com/paper/2308.10741/) (2023)

## Built on by (most cited first, 30 of 172)

- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts](https://mlsanity.com/paper/2311.05608/) (2023)
- [A StrongREJECT for Empty Jailbreaks](https://mlsanity.com/paper/2402.10260/) (2024)
- [Improving Alignment and Robustness with Circuit Breakers](https://mlsanity.com/paper/2406.04313/) (2024)
- [Survey of Vulnerabilities in Large Language Models Revealed by Adversarial Attacks](https://mlsanity.com/paper/2310.10844/) (2023)
- [How Robust is Google's Bard to Adversarial Image Attacks?](https://mlsanity.com/paper/2309.11751/) (2023)
- [Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast](https://mlsanity.com/paper/2402.08567/) (2024)
- [Jailbreaking Attack against Multimodal Large Language Model](https://mlsanity.com/paper/2402.02309/) (2024)
- [Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems](https://mlsanity.com/paper/2504.01990/) (2025)
- [Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation](https://mlsanity.com/paper/2403.09572/) (2024)
- [Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models](https://mlsanity.com/paper/2402.12336/) (2024)
- [Dissecting Adversarial Robustness of Multimodal LM Agents](https://mlsanity.com/paper/2406.12814/) (2024)
- [Coercing LLMs to do and reveal (almost) anything](https://mlsanity.com/paper/2402.14020/) (2024)
- [Defending Against Unforeseen Failure Modes with Latent Adversarial Training](https://mlsanity.com/paper/2403.05030/) (2024)
- [SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model](https://mlsanity.com/paper/2406.12030/) (2024)
- [A Survey of Attacks on Large Vision-Language Models: Resources, Advances, and Future Trends](https://mlsanity.com/paper/2407.07403/) (2024)
- [Red-Teaming for Generative AI: Silver Bullet or Security Theater?](https://mlsanity.com/paper/2401.15897/) (2024)
- [PsySafe: A Comprehensive Framework for Psychological-based Attack, Defense, and Evaluation of Multi-agent System Safety](https://mlsanity.com/paper/2401.11880/) (2024)
- [Visibility into AI Agents](https://mlsanity.com/paper/2401.13138/) (2024)
- [Best-of-N Jailbreaking](https://mlsanity.com/paper/2412.03556/) (2024)
- [Safety of Multimodal Large Language Models on Images and Texts](https://mlsanity.com/paper/2402.00357/) (2024)
- [JailbreakZoo: Survey, Landscapes, and Horizons in Jailbreaking Large Language and Vision-Language Models](https://mlsanity.com/paper/2407.01599/) (2024)
- [Exploring the Frontier of Vision-Language Models: A Survey of Current Methodologies and Future Directions](https://mlsanity.com/paper/2404.07214/) (2024)
- [Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety](https://mlsanity.com/paper/2502.05206/) (2025)
- [Jailbreaking GPT-4V via Self-Adversarial Attacks with System Prompts](https://mlsanity.com/paper/2311.09127/) (2023)
- [Query-Based Adversarial Prompt Generation](https://mlsanity.com/paper/2402.12329/) (2024)
- [ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time](https://mlsanity.com/paper/2410.06625/) (2024)
- [White-box Multimodal Jailbreaks Against Large Vision-Language Models](https://mlsanity.com/paper/2405.17894/) (2024)
- [Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency](https://mlsanity.com/paper/2501.04931/) (2025)
- [Shadowcast: Stealthy Data Poisoning Attacks Against Vision-Language Models](https://mlsanity.com/paper/2402.06659/) (2024)

## Abstract (arXiv)

Are foundation models secure against malicious actors? In this work, we focus on the image input to a vision-language model (VLM). We discover image hijacks, adversarial images that control the behaviour of VLMs at inference time, and introduce the general Behaviour Matching algorithm for training image hijacks. From this, we derive the Prompt Matching method, allowing us to train hijacks matching the behaviour of an arbitrary user-defined text prompt (e.g. 'the Eiffel Tower is now located in Rome') using a generic, off-the-shelf dataset unrelated to our choice of prompt. We use Behaviour Matching to craft hijacks for four types of attack, forcing VLMs to generate outputs of the adversary's choice, leak information from their context window, override their safety training, and believe false statements. We study these attacks against LLaVA, a state-of-the-art VLM based on CLIP and LLaMA-2, and find that all attack types achieve a success rate of over 80%. Moreover, our attacks are automated and require only small image perturbations.