Unified autoregressive models (UAMs) generate both text and image tokens in a single pass, simplifying multimodal generation but introducing new vulnerabilities. The authors present the Token by Token Backdoor Attack (ToBAC), the first backdoor attack targeting UAMs, which can be executed via data-based or model-based poisoning. ToBAC can manipulate both visual outputs and accompanying text, enabling subtle triggers such as the word "cool" to induce brand promotion or ideological influence in a majority of generations, with success rates up to 63.1% against JanusPro.
arXiv:2511.18921v2 Announce Type: replace Abstract: Backdoor attacks undermine the reliability and trustworthiness of machine learning systems by injecting hidden behaviors that can be maliciously ac...
arXiv:2604. 05809v2 Announce Type: replace-cross Abstract: This paper presents Text-Guided Backdoor (TGB), an adjustable backdoor attack against multimodal pretrained models that uses natural-word triggers, namely words that can naturally occur in ordinary textual inputs.
Vis-Poison is a novel attack that poisons multimodal retrieval-augmented generation systems by inserting attacker-controlled images as visual evidence, without altering any textual metadata. The attack uses an automated multi-agent approach to create visually plausible poisoned images and has been tested on two multimodal RAG pipelines, four embedding models, and six generation models. In black-box settings, Vis-Poison achieves an end-to-end success rate between 40.16% and 65.40% against 30,000-entry knowledge bases, and remains effective against various multimodal large language models with an average success rate above 60%.
The paper investigates how multimodal large language models are more susceptible to prompt injection when adversarial instructions are presented as text rather than as non-textual inputs like images. It proposes a training‑free defense that renders untrusted payloads into typographic images (or audio) before they reach the model, a method called Pictionary. Experiments on ten models and two benchmarks show that this approach significantly lowers attack success rates while maintaining normal functionality, and that fine‑tuning on image‑rendered instructions can further reduce the modality gap.
The paper examines how the tight integration of multimodal understanding and image generation in Unified Multimodal Models (UMMs) can create new safety vulnerabilities. It introduces RICE, an attack framework that exploits bidirectional interactions between generation and understanding to propagate unsafe signals across modalities. Experiments demonstrate high attack success rates in both Generation‑to‑Understanding and Understanding‑to‑Generation pathways, revealing previously overlooked safety weaknesses in UMMs.
arXiv:2606. 02947v1 Announce Type: new Abstract: Supervised fine-tuning is the predominant approach for adapting autoregressive vision-language models to downstream tasks.
arXiv:2511.14301v4 Announce Type: replace-cross Abstract: Transformer-based models are highly susceptible to backdoor attacks via supervised fine-tuning (SFT). To red-team existing data-poisoning def...
arXiv:2607. 15657v1 Announce Type: cross Abstract: Multimodal AI agents increasingly rely on persistent long-term memory to ground generation in past visual and textual episodes.
DiSCO is a zero‑shot, black‑box defense for text‑to‑image models that operates solely at the prompt level. It expands prompts with a distribution‑guided suffix using beam search and contrastive scoring against safe and unsafe image pools generated by the target model, iteratively refining until safe content is produced. The method improves safety on the I2P benchmark under various red‑teaming attacks, reducing attack success rates by 37.7% and 25.13% while preserving semantic fidelity and image coherence.
arXiv:2606. 09700v1 Announce Type: cross Abstract: Large language model (LLM)-powered content moderation systems have become a critical defense against harmful online content.
arXiv:2610. 01871v1 Announce Type: cross Abstract: Multimodal Retrieval-Augmented Generation (MRAG) has emerged as a reliable and cost-effective technique of grounding the generative capabilities of Multimodal Large Language Models (MLLMs) into relevant, up-to-date, external knowledge.
Text-to-image diffusion models have achieved high visual fidelity and broad adoption, but remain vulnerable to safety violations when adversaries exploit them to synthesize illicit content. Existing alignment paradigms, from input sanitization to structural feature pruning, are largely organized around unsafe concepts explicitly exposed during filtering, editing, or localization.