Abstract
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.
Abstract (translated)
多模态大型语言模型（MLLMs）中的对象幻觉是一种持续存在的失败模式，导致模型在图像中感知到实际不存在的对象。目前，研究这种模型弱点的方法主要是通过使用静态基准测试来完成的，这些基准测试包含固定的视觉场景，这可能无法揭示特定于某个模型或未被预见的幻觉漏洞。为此，我们引入了一种名为GHOST（通过优化隐蔽标记生成幻觉）的新方法，旨在通过主动生成诱导幻觉的图像对MLLMs进行压力测试。这种方法是全自动化的，并不需要人类监督或先验知识。 GHOST的工作原理是在不包含目标对象的情况下，在图像嵌入空间中进行优化以误导模型，然后引导基于该嵌入条件的扩散模型生成看起来自然的图像。生成的图像在视觉上仍然保持与原始输入相似和高质量的状态，但会引入一些微妙的误导性线索，导致模型产生幻觉。 我们通过一系列包括推理模型（如GLM-4.1V-Thinking）在内的不同模型对我们的方法进行了评估，并且达到了超过28%的幻觉成功率，相比之下，先前基于数据驱动的方法仅为大约1%。我们还通过定量指标和人类评价确认了生成的图像是高质量且无对象存在的。 此外，GHOST揭示了一种可迁移性漏洞：为Qwen2.5-VL优化的图像在GPT-4o中诱导幻觉的成功率为66.5%。最后，我们展示了对我们的图片进行微调可以减轻幻觉现象的发生率，从而将GHOST定位为一种诊断和纠正工具，有助于构建更可靠的多模态系统。
URL
https://arxiv.org/abs/2509.25178
