# GHOST: Hallucination-Inducing Image Generation for Multimodal LLMs

arXiv: https://arxiv.org/abs/2509.25178 · HTML: https://mlsanity.com/paper/2509.25178/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Hallucination mitigation in LVLMs, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 43 of its 57 references that are in the map.
- First posted on arXiv 29 Sep 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Hallucination mitigation in LVLMs](https://mlsanity.com/mllm/hallucination-mitigation-in-lvlms/)

## Builds on (most cited first, 30 of 43)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Deep Unsupervised Learning using Nonequilibrium Thermodynamics](https://mlsanity.com/paper/1503.03585/) (2015)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection](https://mlsanity.com/paper/2303.05499/) (2023)
- [Evaluating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2305.10355/) (2023)
- [Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization](https://mlsanity.com/paper/1911.08731/) (2019)
- [Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs](https://mlsanity.com/paper/2401.06209/) (2024)
- [YOLO-World: Real-Time Open-Vocabulary Object Detection](https://mlsanity.com/paper/2401.17270/) (2024)
- [GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning](https://mlsanity.com/paper/2507.01006/) (2025)
- [Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations](https://mlsanity.com/paper/2204.02937/) (2022)
- [AMBER: An LLM-free Multi-dimensional Benchmark for MLLMs Hallucination Evaluation](https://mlsanity.com/paper/2311.07397/) (2023)
- [On Evaluating Adversarial Robustness of Large Vision-Language Models](https://mlsanity.com/paper/2305.16934/) (2023)
- [Negative Object Presence Evaluation (NOPE) to Measure Object Hallucination in Vision-Language Models](https://mlsanity.com/paper/2310.05338/) (2023)
- [Content-based Unrestricted Adversarial Attack](https://mlsanity.com/paper/2305.10665/) (2023)
- [CIEM: Contrastive Instruction Evaluation Method for Better Instruction Tuning](https://mlsanity.com/paper/2309.02301/) (2023)
- [CF-VLM:CounterFactual Vision-Language Fine-tuning](https://mlsanity.com/paper/2506.17267/) (2025)
- [The Curse of Multi-Modalities: Evaluating Hallucinations of Large Multimodal Models across Language, Visual, and Audio](https://mlsanity.com/paper/2410.12787/) (2024)
- [AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models](https://mlsanity.com/paper/2410.05346/) (2024)
- [Exposing and Mitigating Spurious Correlations for Cross-Modal Retrieval](https://mlsanity.com/paper/2304.03391/) (2023)
- [Mass-Producing Failures of Multimodal Systems with Language Models](https://mlsanity.com/paper/2306.12105/) (2023)
- [MM-SpuBench: Towards Better Understanding of Spurious Biases in Multimodal LLMs](https://mlsanity.com/paper/2406.17126/) (2024)
- [Aya Vision: Advancing the Frontier of Multilingual Multimodality](https://mlsanity.com/paper/2505.08751/) (2025)

## Built on by (most cited first, 4 of 4)

- [S2-FracMix: Label-Preserving Self-Saliency Mixup Augmentation](https://mlsanity.com/paper/2606.25784/) (2026)
- [InstructMixup: Instruction-Guided Salient Patch Editing for Robust Data Augmentation](https://mlsanity.com/paper/2607.19324/) (2026)
- [From Pixel Generation to Topological Inference: Structural Dual Super-Resolution for Trustworthy Cross-Physical-Domain Trabecular Morphology Learning](https://mlsanity.com/paper/2609.34716/) (2026)
- [MEND: Label-Free Detection, Localisation, and Correction of Latent Hallucination in World Models](https://mlsanity.com/paper/2609.39182/) (2026)

## Abstract (arXiv)

Object hallucination in Multimodal Large Language Models (MLLMs) is a persistent failure mode that causes the model to perceive objects absent in the image. This weakness of MLLMs is currently studied using static benchmarks with fixed visual scenarios, which preempts the possibility of uncovering model-specific or unanticipated hallucination vulnerabilities. We introduce GHOST (Generating Hallucinations via Optimizing Stealth Tokens), a method designed to stress-test MLLMs by actively generating images that induce hallucination. GHOST is fully automatic and requires no human supervision or prior knowledge. It operates by optimizing in the image embedding space to mislead the model while keeping the target object absent, and then guiding a diffusion model conditioned on the embedding to generate natural-looking images. The resulting images remain visually natural and close to the original input, yet introduce subtle misleading cues that cause the model to hallucinate. We evaluate our method across a range of models, including reasoning models like GLM-4.1V-Thinking, and achieve a hallucination success rate exceeding 28%, compared to around 1% in prior data-driven discovery methods. We confirm that the generated images are both high-quality and object-free through quantitative metrics and human evaluation. Also, GHOST uncovers transferable vulnerabilities: images optimized for Qwen2.5-VL induce hallucinations in GPT-4o at a 66.5% rate. Finally, we show that fine-tuning on our images mitigates hallucination, positioning GHOST as both a diagnostic and corrective tool for building more reliable multimodal systems.