Multimodal Large Language Models (MLLMs) have achieved remarkable performance across vision-language tasks. Recent advancements allow these models to process multiple images as inputs. However, the vulnerabilities of multi-image MLLMs remain unexplored. Existing adversarial attacks focus on single-image settings and often assume a white-box threat model which is impractical in many real-world scenarios. This paper introduces LAMP, a black-box method for learning UAPs targeting multi-image MLLMs. LAMP applies an attention-based constraint that which prevents the model from effectively aggregating information across images. LAMP also introduces a novel cross-image contagious constraint that forces perturbed tokens to influence clean tokens to spread adversarial effects without requiring all inputs to be modified. Additionally, an index-attention suppression loss creates a robust position invariant attack. Experimental results show that LAMP outperforms SOTA baselines and achieves the highest attack success rates across multiple vision-language tasks.

Aafiya Shamshad Hussain, Gaurav Srivastava, Alvi Md Ishmam, Zaber Ibn Abdul Hakim, Chris Thomas. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

In recent years there has been enormous interest in vision-language models trained using self-supervised objectives. However, the use of large-scale datasets scraped from the web for training also makes these models vulnerable to potential security threats, such as backdooring and poisoning attacks. In this paper, we propose a method for mitigating such attacks on contrastively trained vision-language models. Our approach leverages external knowledge extracted from a language model to prevent models from learning correlations between image regions which lack strong alignment with external knowledge. We do this by imposing constraints to enforce that attention paid by the model to visual regions is proportional to the alignment of those regions with external knowledge. We conduct extensive experiments using a variety of recent backdooring and poisoning attacks on multiple datasets and architectures. Our results clearly demonstrate that our proposed approach is highly effective at defending against such attacks across multiple settings, while maintaining model utility and without requiring any changes at inference time.

Validating claims from misinformation is a highly challenging task that involves understanding how each factual assertion within the claim relates to a set of trusted source materials. Existing approaches often make coarse-grained predictions but fail to identify the specific aspects of the claim that are troublesome and the specific evidence relied upon. In this paper, we introduce a method and new benchmark for this challenging task. Our method predicts the fine-grained logical relationship of each aspect of the claim from a set of multimodal documents, which include text, image(s), video(s), and audio(s). We also introduce a new benchmark (M^3DC) of claims requiring multimodal multidocument reasoning, which we construct using a novel claim synthesis technique. Experiments show that our approach significantly outperforms state-of-the-art baselines on this challenging task on two benchmarks while providing finer-grained predictions, explanations, and evidence.

Existing vision-language understanding benchmarks largely consist of images of objects in their usual contexts.As a consequence, recent multimodal large language models can perform well with only a shallow visual understanding by relying on background language biases. Thus, strong performance on these benchmarks does not necessarily correlate with strong visual understanding. In this paper, we release JourneyBench, a comprehensive human-annotated benchmark of generated images designed to assess the model's fine-grained multimodal reasoning abilities across five tasks: complementary multimodal chain of thought, multi-image VQA, imaginary image captioning, VQA with hallucination triggers, and fine-grained retrieval with sample-specific distractors.Unlike existing benchmarks, JourneyBench explicitly requires fine-grained multimodal reasoning in unusual imaginary scenarios where language bias and holistic image gist are insufficient. We benchmark state-of-the-art models on JourneyBench and analyze performance along a number of fine-grained dimensions. Results across all five tasks show that JourneyBench is exceptionally challenging for even the best models, indicating that models' visual reasoning abilities are not as strong as they first appear. We discuss the implications of our findings and propose avenues for further research.

Online hateful speech detection and classification in social media for the various major languages other than English has drawn the attention of researchers recently. In this paper, we develop Machine Learning (ML) algorithms based model, as well as Gated Recurrent Unit (GRU), based deep neural network model for classifying users' comments on Facebook pages. We have collected, annotated 5,126 Bengali comments and classified them into six classes - Hate Speech, Communal Attack, Inciteful, Religious Hatred, Political Comments, and Religious Comments. The produced corpus is the first contribution to the field of hateful speech detection in the Bengali language for social media. Finally, we employ several machine learning algorithms, compare the performance, and attained 52.20% accuracy in Random Forest. The accuracy is improved in the case of GRU based model (70.10% accuracy) about 18%.