Retrieval-augmented generation (RAG) has emerged as a powerful paradigm for enhancing multimodal large language models by grounding their responses in external, factual knowledge and thus mitigating hallucinations. However, the integration of externally sourced knowledge bases introduces a critical attack surface. Adversaries can inject malicious multimodal content capable of influencing both retrieval and downstream generation. In this work, we present MM-MEPA, a multimodal poisoning attack that targets the metadata components of image-text entries while leaving the associated visual content unaltered. By only manipulating the metadata, MM-MEPA can still steer multimodal retrieval and induce attacker-desired model responses. We evaluate the attack across multiple benchmark settings and demonstrate its severity. MM-MEPA achieves an attack success rate of up to 91\% consistently disrupting system behaviors across four retrievers and two multimodal generators. Additionally, we assess representative defense strategies and find them largely ineffective against this form of metadata-only poisoning. Our findings expose a critical vulnerability in multimodal RAG and underscore the urgent need for more robust, defense-aware retrieval and knowledge integration methods.

Vision-Language Models (VLMs) are increasingly deployed in safety-critical applications, making their adversarial robustness a crucial concern. While adversarial knowledge distillation has shown promise in transferring robustness from teacher to student models, traditional single-teacher approaches suffer from limited knowledge diversity, slow convergence, and difficulty in balancing robustness and accuracy. To address these challenges, we propose MMT-ARD: a Multimodal Multi-Teacher Adversarial Robust Distillation framework. Our key innovation is a dual-teacher knowledge fusion architecture that collaboratively optimizes clean feature preservation and robust feature enhancement. To better handle challenging adversarial examples, we introduce a dynamic weight allocation strategy based on teacher confidence, enabling adaptive focus on harder samples. Moreover, to mitigate bias among teachers, we design an adaptive sigmoid-based weighting function that balances the strength of knowledge transfer across modalities. Extensive experiments on ImageNet and zero-shot benchmarks demonstrate that MMT-ARD improves robust accuracy by +4.32% and zero-shot accuracy by +3.5% on the ViT-B-32 model, while achieving a 2.3x increase in training efficiency over traditional single-teacher methods. These results highlight the effectiveness and scalability of MMT-ARD in enhancing the adversarial robustness of multimodal large models. Our codes are available at https://github.com/itsnotacie/MMT-ARD.

Retrieval-Augmented Generation (RAG) has emerged as a powerful approach to boost the capabilities of large language models (LLMs) by incorporating external, up-to-date knowledge sources. However, this introduces a potential vulnerability to knowledge poisoning attacks, where attackers can compromise the knowledge source to mislead the generation model. One such attack is the PoisonedRAG in which the injected adversarial texts steer the model to generate an attacker-chosen response to a target question. In this work, we propose novel defense methods, FilterRAG and ML-FilterRAG, to mitigate the PoisonedRAG attack. First, we propose a new property to uncover distinct properties to differentiate between adversarial and clean texts in the knowledge data source. Next, we employ this property to filter out adversarial texts from clean ones in the design of our proposed approaches. Evaluation of these methods using benchmark datasets demonstrate their effectiveness, with performances close to those of the original RAG systems.