Jailbreak prompts can bypass alignment guardrails in large language models (LLMs) and elicit unsafe outputs, making reliable deployment-time detection critical. Prior detection approaches largely rely on a fixed metric space, e.g., raw inputs, gradients, or hidden features, in which benign and jailbreak prompts are linearly separable. We show this assumption breaks under (i) pseudo-malicious prompts that are benign by intent but contain safety-related keywords, and (ii) adaptive attacks that explicitly optimize against the deployed detector. To overcome this limitation, we shift our focus from identifying a universal metric space to analyzing the more robust neighborhood structure of the underlying data manifold. We present Manifold Trajectory Kinetics (MTK), which treats an LLM as a kinetic system transforming inputs into outputs and detects jailbreaks by tracking how a prompt's neighborhood structure evolves across layers. Benign prompts remain close to benign neighborhoods throughout inference, whereas jailbreak prompts exhibit a characteristic trajectory that begins near malicious seeds and later strategically shifts toward benign neighborhoods to evade refusal.Across four LLMs and ten jailbreak attacks, MTK achieves strong robustness to both failure modes: on pseudo-malicious prompts, it attains a jailbreak true positive rate of 95

Recent commercial image-generation models can generate high-quality images with readable text (e.g., posters, infographics, and manuals), attracting considerable attention. Yet we first show that this same capability also introduces a previously unreported safety vulnerability: these systems may refuse to generate harmful text directly, yet permit the same content when rendered as text within generated images, i.e., safety alignment does not reliably transfer from textual outputs to text embedded in images. In this paper, unlike existing visual jailbreaks against image-generation models, which primarily induce models to generate harmful visual objects or scenes, we introduce the concept of instruction-dense visual jailbreaks, in which image-generation models produce detailed, readable, and actionable harmful instructions within images. Such outputs can amplify harm because the rendered instructions can be readily read and widely spread. To instantiate this threat, we propose TYPO, a black-box framework that exploits this safety gap by automatically generating adversarial TYPOgraphy prompts, which covertly steer image-generation models to express harmful intent as highly legible, typographically structured text. Specifically, TYPO decomposes prompt generation into two channels: a textual channel for reframing the target intent, and a visual channel for specifying its presentation form. We formulate these two channels as a dual-channel textual-visual strategy space and optimize candidate strategy combinations through an adaptive combinatorial search. Extensive experiments across four commercial models (i.e., GPT-Image-2, Nano Banana Pro, Qwen-Image-2, and Seedream 5.0 Lite) show that TYPO substantially outperforms nine representative jailbreak attacks by 50.2

Unlearnable examples (UEs) aim to compromise model training by injecting imperceptible perturbations to clean samples. However, existing UE schemes exhibit limited robustness against advanced defenses due to their heuristic design or narrowly scoped domain perturbations. To address this, we propose \texttt{DUNE}, a \underline{\textbf{D}}ual-branch \underline{\textbf{UN}}learnable \underline{\textbf{E}}nsemble perturbation optimization approach. Specifically, \texttt{DUNE} separately optimizes perturbations in the spatial and color domains to establish the mapping between perturbations and shift-induced labels. This design extends the perturbation domain to increase noise intensity for improving robustness and drives the models to learn perturbation-oriented features with degraded generalization, thereby achieving unlearnability. To strengthen \texttt{DUNE}'s performance, we further propose an unlearnability-enhancing ensemble strategy that aggregates diverse pre-trained models during the dual-branch optimization. Extensive experiments on benchmark datasets CIFAR-10 and ImageNet verify that \texttt{DUNE}'s robustness outperforms 12 SOTA UE schemes under 7 mainstream defenses, yielding a lower average test accuracy of 14.95\% to 50.82\%.

Federated learning (FL) is one of the most widely used distributed machine learning frameworks. However, FL is susceptible to poisoning attacks that can degrade the quality of the global model. Recent studies on fine-grained poisoning attacks highlight a strategic shift where attackers no longer prioritize maximal disruption of the global model, but instead control the degree of model poisoning to maintain stealth and avoid detection. However, research on fine-grained poisoning is still in the infant stage. Numerous fundamental questions have yet to be addressed, including its underlying mechanisms and optimization strategies. To this end, we introduces FGP, the first comprehensive framework for Fine-Grained Poisoning on FL, which allows adversaries to precisely manipulate the global model by strategically inducing accurate and stealthy sub-optimal solution. Fundamentally, FGP innovatively formalizes fine-grained attacks as an optimization problem to minimize the distance between the current global model and the adversary's target (a sub-optimal solution). It then employs a real-time search strategy to dynamically refine the malicious model updates in each round. To ensure optimal attack performance, we further introduce a novel topology-based approach as the error feedback. Additionally, we present a formal convergence analysis of our attacks. Armed with FGP, we conduct a comprehensive evaluation of FL's robustness against fine-grained poisoning across diverse settings. Results demonstrate that FGP significantly outperforms the prior work, achieving an average 6.5x higher attack accuracy.

Diffusion-based image-to-video (I2V) generation has become a central direction in generative models by turning a reference image, with optional conditions, into a temporally coherent video. Compared with broader video generation settings, this task places stricter demands on content consistency, identity preservation, and motion coherence. Although the literature grows rapidly, existing works mostly discuss I2V generation within broader topics and still lack a dedicated taxonomy together with a systematic analysis centered on this field. This work addresses that gap by treating diffusion I2V generation as a standalone subject. It first reviews the task formulation, model architectures, datasets, and evaluation metrics, and then organizes existing methods through a taxonomy based on architecture and training paradigm. It further distills four core designs, namely condition encoding, temporal modeling, noise prior design, and spatial-temporal upsampling, and discusses representative application scenarios together with major open challenges.

Segmentation models exhibit significant vulnerability to adversarial examples in white-box settings, but existing adversarial attack methods often show poor transferability across different segmentation models. While some researchers have explored transfer-based adversarial attack (i.e., transfer attack) methods for segmentation models, the complex contextual dependencies within these models and the feature distribution gaps between surrogate and target models result in unsatisfactory transfer success rates. To address these issues, we propose SegTrans, a novel transfer attack framework that divides the input sample into multiple local regions and remaps their semantic information to generate diverse enhanced samples. These enhanced samples replace the original ones for perturbation optimization, thereby improving the transferability of adversarial examples across different segmentation models. Unlike existing methods, SegTrans only retains local semantic information from the original input, rather than using global semantic information to optimize perturbations. Extensive experiments on two benchmark datasets, PASCAL VOC and Cityscapes, four different segmentation models, and three backbone networks show that SegTrans significantly improves adversarial transfer success rates without introducing additional computational overhead. Compared to the current state-of-the-art methods, SegTrans achieves an average increase of 8.55

Vision-Language Models (VLMs) are known to be vulnerable to adversarial attacks, where subtle perturbations to images or texts induce erroneous outputs. However, most text-based attacks are adapted from language-model-centric methods, in which the visual input is fixed during optimization, resulting in adversarial prompts that are tied to specific images and thus limiting their attack effectiveness. To this end, we first introduce a new research perspective: cross-image transferability for adversarial prompts. We then propose GhostPrompt, an adversarial prompt that is optimized once and reused to steer VLM outputs toward attacker-specified responses across diverse images. GhostPrompt employs a joint optimization that distills image-invariant adversarial features into the prompt by "worst-case" generation. Specifically, it alternates between constructing hard visual conditions for the current prompt and updating the prompt to remain effective under these conditions. Extensive experiments on prevalent VLMs verify that achieves an improvement of over 30

Benefiting from its superior feature learning capabilities and efficiency, deep hashing has achieved remarkable success in large-scale image retrieval. Recent studies have demonstrated the vulnerability of deep hashing models to backdoor attacks. Although these studies have shown promising attack results, they rely on access to the training dataset to implant the backdoor. In the real world, obtaining such data (e.g., identity information) is often prohibited due to privacy protection and intellectual property concerns. Embedding backdoors into deep hashing models without access to the training data, while maintaining retrieval accuracy for the original task, presents a novel and challenging problem. In this paper, we propose DarkHash, the first data-free backdoor attack against deep hashing. Specifically, we design a novel shadow backdoor attack framework with dual-semantic guidance. It embeds backdoor functionality and maintains original retrieval accuracy by fine-tuning only specific layers of the victim model using a surrogate dataset. We consider leveraging the relationship between individual samples and their neighbors to enhance backdoor attacks during training. By designing a topological alignment loss, we optimize both individual and neighboring poisoned samples toward the target sample, further enhancing the attack capability. Experimental results on four image datasets, five model architectures, and two hashing methods demonstrate the high effectiveness of DarkHash, outperforming existing state-of-the-art backdoor attack methods. Defense experiments show that DarkHash can withstand existing mainstream backdoor defense methods.

Object detection models are vulnerable to backdoor attacks, where attackers poison a small subset of training samples by embedding a predefined trigger to manipulate prediction. Detecting poisoned samples (i.e., those containing triggers) at test time can prevent backdoor activation. However, unlike image classification tasks, the unique characteristics of object detection -- particularly its output of numerous objects -- pose fresh challenges for backdoor detection. The complex attack effects (e.g., "ghost" object emergence or "vanishing" object) further render current defenses fundamentally inadequate. To this end, we design TRAnsformation Consistency Evaluation (TRACE), a brand-new method for detecting poisoned samples at test time in object detection. Our journey begins with two intriguing observations: (1) poisoned samples exhibit significantly more consistent detection results than clean ones across varied backgrounds. (2) clean samples show higher detection consistency when introduced to different focal information. Based on these phenomena, TRACE applies foreground and background transformations to each test sample, then assesses transformation consistency by calculating the variance in objects confidences. TRACE achieves black-box, universal backdoor detection, with extensive experiments showing a 30% improvement in AUROC over state-of-the-art defenses and resistance to adaptive attacks.

Embodied AI represents systems where AI is integrated into physical entities, enabling them to perceive and interact with their surroundings. Large Language Model (LLM), which exhibits powerful language understanding abilities, has been extensively employed in embodied AI by facilitating sophisticated task planning. However, a critical safety issue remains overlooked: could these embodied LLMs perpetrate harmful behaviors? In response, we introduce BadRobot, a novel attack paradigm aiming to make embodied LLMs violate safety and ethical constraints through typical voice-based user-system interactions. Specifically, three vulnerabilities are exploited to achieve this type of attack: (i) manipulation of LLMs within robotic systems, (ii) misalignment between linguistic outputs and physical actions, and (iii) unintentional hazardous behaviors caused by world knowledge's flaws. Furthermore, we construct a benchmark of various malicious physical action queries to evaluate BadRobot's attack performance. Based on this benchmark, extensive experiments against existing prominent embodied LLM frameworks (e.g., Voxposer, Code as Policies, and ProgPrompt) demonstrate the effectiveness of our BadRobot. More demonstrations are available at an anonymous address: https://Embodied-LLMs-Safety.github.io. Warning: This paper contains harmful AI-generated language and aggressive actions.

Diffusion models has threatened image authenticity by enabling highly realistic fakes. Proactive defense offers protection by adding a "protective layer" that resists such manipulation. However, current proactive defenses mainly focus on text-guided editing but are less effective for the more challenging exemplar-guided tasks. To bridge this gap, we propose Proactive Spatial-Frequency Defense (PSFD), a novel proactive defense for exemplar-guided image editing. PSFD leverages adversarial attack to add subtle perturbations to make images immune to editing. We apply protections in both the frequency and spatial domains. Spatial perturbation disrupts feature extraction by forcing visual encoders to map the image to "bad" representations. Frequency perturbation tweaks the high-frequency components to distort the image’s texture information. We design two optimization strategies: PSFD-U that aims to generate maximal variation and PSFD-T that seeks to achieve specific editing styles. Extensive experiments on MS-COCO and ImageNet demonstrate PSFD’s strong defensive capabilities and transferability.

As deep neural networks (DNNs) are widely applied in the physical world, many researches are focusing on physical-world adversarial examples (PAEs), which introduce perturbations to inputs and cause the model's incorrect outputs. However, existing PAEs face two challenges: unsatisfactory attack performance (i.e., poor transferability and insufficient robustness to environment conditions), and difficulty in balancing attack effectiveness with stealthiness, where better attack effectiveness often makes PAEs more perceptible. In this paper, we explore a novel perturbation-based method to overcome the challenges. For the first challenge, we introduce a strategy Deceptive RF injection based on robust features (RFs) that are predictive, robust to perturbations, and consistent across different models. Specifically, it improves the transferability and robustness of PAEs by covering RFs of other classes onto the predictive features in clean images. For the second challenge, we introduce another strategy Adversarial Semantic Pattern Minimization, which removes most perturbations and retains only essential adversarial patterns in AEsBased on the two strategies, we design our method Robust Feature Coverage Attack (RFCoA), comprising Robust Feature Disentanglement and Adversarial Feature Fusion. In the first stage, we extract target class RFs in feature space. In the second stage, we use attention-based feature fusion to overlay these RFs onto predictive features of clean images and remove unnecessary perturbations. Experiments show our method's superior transferability, robustness, and stealthiness compared to existing state-of-the-art methods. Additionally, our method's effectiveness can extend to Large Vision-Language Models (LVLMs), indicating its potential applicability to more complex tasks.

Vision-Language Models (VLMs), with their strong reasoning and planning capabilities, are widely used in embodied decision-making (EDM) tasks in embodied agents, such as autonomous driving and robotic manipulation. Recent research has increasingly explored adversarial attacks on VLMs to reveal their vulnerabilities. However, these attacks either rely on overly strong assumptions, requiring full knowledge of the victim VLM, which is impractical for attacking VLM-based agents, or exhibit limited effectiveness. The latter stems from disrupting most semantic information in the image, which leads to a misalignment between the perception and the task context defined by system prompts. This inconsistency interrupts the VLM's reasoning process, resulting in invalid outputs that fail to affect interactions in the physical world. To this end, we propose a fine-grained adversarial attack framework, ADVEDM, which modifies the VLM's perception of only a few key objects while preserving the semantics of the remaining regions. This attack effectively reduces conflicts with the task context, making VLMs output valid but incorrect decisions and affecting the actions of agents, thus posing a more substantial safety threat in the physical world. We design two variants of based on this framework, ADVEDM-R and ADVEDM-A, which respectively remove the semantics of a specific object from the image and add the semantics of a new object into the image. The experimental results in both general scenarios and EDM tasks demonstrate fine-grained control and excellent attack performance.

With the rapid advancement of deep learning, the model robustness has become a significant research hotspot, i.e., adversarial attacks on deep neural networks. Existing works primarily focus on image classification tasks, aiming to alter the model’s predicted labels. Due to the output complexity and deeper network architectures, research on adversarial examples for segmentation models is still limited, particularly for universal adversarial perturbations. In this paper, we propose a novel universal adversarial attack method designed for segmentation models, which includes dual feature separation and low-frequency scattering modules. The two modules guide the training of adversarial examples in the pixel and frequency space, respectively. Experiments demonstrate that our method achieves high attack success rates surpassing the state-of-the-art methods, and exhibits strong transferability across different models.

Robotic manipulation policies are increasingly empowered by large language models (LLMs) and vision-language models (VLMs), leveraging their understanding and perception capabilities. Recently, inference-time attacks against robotic manipulation have been extensively studied, yet backdoor attacks targeting model supply chain security in robotic policies remain largely unexplored. To fill this gap, we propose , a backdoor injection framework for model supply chain attack scenarios, which embeds a malicious module into modular robotic policies via backdoor relationships to manipulate the LLM-to-VLM pathway and compromise the system. Our vanilla design instantiates this module as a backdoor-finetuned VLM. To further enhance attack performance, we propose a prime scheme by introducing the concept of LVLM-as-a-backdoor, which leverages in-context instruction learning (ICIL) to steer large vision-language model (LVLM) behavior through backdoored system prompts. Moreover, we develop three types of prime attacks, permutation, stagnation, and intentional, achieving flexible backdoor attack effects. Extensive physical-world and simulator experiments on 18 real-world manipulation tasks and 4 VLMs verify the superiority of proposed

Object detection tasks, crucial in safety-critical systems like autonomous driving, focus on pinpointing object locations. These detectors are known to be susceptible to backdoor attacks. However, existing backdoor techniques have primarily been adapted from classification tasks, overlooking deeper vulnerabilities specific to object detection. This paper is dedicated to bridging this gap by introducing Detector Collapse} (DC), a brand-new backdoor attack paradigm tailored for object detection. DC is designed to instantly incapacitate detectors (i.e., severely impairing detector's performance and culminating in a denial-of-service). To this end, we develop two innovative attack schemes: Sponge for triggering widespread misidentifications and Blinding for rendering objects invisible. Remarkably, we introduce a novel poisoning strategy exploiting natural objects, enabling DC to act as a practical backdoor in real-world environments. Our experiments on different detectors across several benchmarks show a significant improvement ($\sim$10\%-60\% absolute and $\sim$2-7$\times$ relative) in attack efficacy over state-of-the-art attacks.

The backdoor attack on deep neural network models implants malicious data patterns in a model to induce attacker-desirable behaviors. Existing defense methods fall into the online and offline categories, in which the offline models achieve state-of-the-art detection rates but are restricted by heavy computation overhead. In contrast, their more deployable online counterparts lack the means to detect source-specific backdoors with large sizes. This work proposes a new online backdoor detection method—Reverse Backdoor Distillation (RBD) to handle issues associated with source-specific and source-agnostic backdoor attacks. RBD, designed with the novel perspective of distilling instead of erasing backdoor knowledge, is a complementary backdoor detection methodology that can be used in conjunction with other online backdoor defenses. Considering the fact that trigger data will cause overwhelming neuron activation while clean data will not, RBD distills backdoor attack pattern knowledge from a suspicious model to create a shadow model, which is subsequently deployed online along with the original model in scope to predict a backdoor attack. We extensively evaluate RBD on several datasets (MNIST, GTSRB, CIFAR-10) with diverse model architectures and trigger patterns. RBD outperforms online benchmarks in all experimental settings. Notably, RBD demonstrates superior capability in detecting source-specific attacks, where comparison methods fail, underscoring the effectiveness of our proposed technique. Moreover, RBD achieves a computational savings of at least 97%.

Traditional unlearnable strategies have been proposed to prevent unauthorized users from training on the 2D image data. With more 3D point cloud data containing sensitivity information, unauthorized usage of this new type data has also become a serious concern. To address this, we propose the first integral unlearnable framework for 3D point clouds including two processes: (i) we propose an unlearnable data protection scheme, involving a class-wise setting established by a category-adaptive allocation strategy and multi-transformations assigned to samples; (ii) we propose a data restoration scheme that utilizes class-wise inverse matrix transformation, thus enabling authorized-only training for unlearnable data. This restoration process is a practical issue overlooked in most existing unlearnable literature, i.e., even authorized users struggle to gain knowledge from 3D unlearnable data. Both theoretical and empirical results (including 6 datasets, 16 models, and 2 tasks) demonstrate the effectiveness of our proposed unlearnable framework. Our code is available at https://github.com/CGCL-codes/UnlearnablePC.