Recent advances in GAN and diffusion models have significantly improved the realism and controllability of facial deepfake manipulation, raising serious concerns regarding privacy, security, and identity misuse. Proactive defenses attempt to counter this threat by injecting adversarial perturbations into images before manipulation takes place. However, existing approaches remain limited in effectiveness due to suboptimal perturbation injection strategies and are typically designed under white-box assumptions, targeting only simple GAN-based attribute editing. These constraints hinder their applicability in practical real-world scenarios. In this paper, we propose AEGIS, the first diffusion-guided paradigm in which the AdvErsarial facial images are Generated for Identity Shielding. We observe that the limited defense capability of existing approaches stems from the peak-clipping constraint, where perturbations are forcibly truncated due to a fixed $L_\infty$-bounded. To overcome this limitation, instead of directly modifying pixels, AEGIS injects adversarial perturbations into the latent space along the DDIM denoising trajectory, thereby decoupling the perturbation magnitude from pixel-level constraints and allowing perturbations to adaptively amplify where most effective. The extensible design of AEGIS allows the defense to be expanded from purely white-box use to also support black-box scenarios through a gradient-estimation strategy. Extensive experiments across GAN and diffusion-based deepfake generators show that AEGIS consistently delivers strong defense effectiveness while maintaining high perceptual quality. In white-box settings, it achieves robust manipulation disruption, whereas in black-box settings, it demonstrates strong cross-model transferability.

Vision-language models remain susceptible to multimodal jailbreaks and over-refusal because safety hinges on both visual evidence and user intent, while many alignment pipelines supervise only the final response. To address this, we present SaFeR-ToolKit, which formalizes safety decision-making as a checkable protocol. Concretely, a planner specifies a persona, a Perception $\to$ Reasoning $\to$ Decision tool set, and a constrained transition graph, while a responder outputs a typed key-value tool trace before the final answer. To make the protocol reliably followed in practice, we train a single policy with a three-stage curriculum (SFT $\to$ DPO $\to$ GRPO), where GRPO directly supervises tool usage beyond answer-level feedback. Our contributions are two-fold: I. Dataset. The first tool-based safety reasoning dataset, comprising 31,654 examples (SFT 6k, DPO 18.6k, GRPO 6k) plus 1k held-out evaluation. II. Experiments. On Qwen2.5-VL, SaFeR-ToolKit significantly improves Safety/Helpfulness/Reasoning Rigor on 3B (29.39/45.04/4.98 $\to$ 84.40/71.13/78.87) and 7B (53.21/52.92/19.26 $\to$ 86.34/80.79/85.34), while preserving general capabilities (3B: 58.67 $\to$ 59.21; 7B: 66.39 $\to$ 66.81). Codes are available at https://github.com/Duebassx/SaFeR_ToolKit.

The widespread use of publicly available pre-trained encoders from self-supervised learning (SSL) has exposed a critical vulnerability: their susceptibility to downstream-agnostic adversarial examples (DAEs), which are crafted without knowledge of the downstream tasks but capable of misleading downstream models. While several defense methods have been explored recently, they rely primarily on task-specific adversarial fine-tuning, which inevitably limits generalizability and causes catastrophic forgetting and deteriorates benign performance. Different with previous works, we propose a more rigorous defense goal that requires only a single tuning for diverse downstream tasks to defend against DAEs and preserve benign performance. To achieve this defense goal, we introduce Zero-Sacrifice Persistent-Robustness Adversarial Defense (ZePAD), which is inspired by the inherent sensitivity of neural networks to data characteristics. Specifically, ZePAD is a dual-branch structure, which consists of a Multi-Pattern Adversarial Enhancement Branch (MPAE-Branch) that uses two adversarially fine-tuned encoders to strengthen adversarial resistance. The Benign Memory Preservation Branch (BMP-Branch) is trained on local data to ensure adversarial robustness does not compromise benign performance. Surprisingly, we find that ZePAD can directly detect DAEs by evaluating branch confidence, without introducing any adversarial exsample identification task during training. Notably, by enriching feature diversity, our method enables a single adversarial fine-tuning to defend against DAEs across downstream tasks, thereby achieving persistent robustness. Extensive experiments on 11 SSL methods and 6 datasets validate its effectiveness. In certain cases, it achieves a 29.20% improvement in benign performance and a 73.86% gain in adversarial robustness, highlighting its zero-sacrifice property.

With the deep integration of facial recognition into online banking, identity verification, and other networked services, achieving effective decoupling of identity information from visual representations during image storage and transmission has become a critical challenge for privacy protection. To address this issue, we propose SIDeR, a Semantic decoupling-driven framework for unrestricted face privacy protection. SIDeR decomposes a facial image into a machine-recognizable identity feature vector and a visually perceptible semantic appearance component. By leveraging semantic-guided recomposition in the latent space of a diffusion model, it generates visually anonymous adversarial faces while maintaining machine-level identity consistency. The framework incorporates momentum-driven unrestricted perturbation optimization and a semantic-visual balancing factor to synthesize multiple visually diverse, highly natural adversarial samples. Furthermore, for authorized access, the protected image can be restored to its original form when the correct password is provided. Extensive experiments on the CelebA-HQ and FFHQ datasets demonstrate that SIDeR achieves a 99% attack success rate in black-box scenarios and outperforms baseline methods by 41.28% in PSNR-based restoration quality.

Concept erasure aims to suppress sensitive content in diffusion models, but recent studies show that erased concepts can still be reawakened, revealing vulnerabilities in erasure methods. Existing reawakening methods mainly rely on prompt-level optimization to manipulate sampling trajectories, neglecting other generative factors, which limits a comprehensive understanding of the underlying dynamics. In this paper, we model the generation process as an implicit function to enable a comprehensive theoretical analysis of multiple factors, including text conditions, model parameters, and latent states. We theoretically show that perturbing each factor can reawaken erased concepts. Building on this insight, we propose a novel concept reawakening method: Latent space Unblocking for concept REawakening (LURE), which reawakens erased concepts by reconstructing the latent space and guiding the sampling trajectory. Specifically, our semantic re-binding mechanism reconstructs the latent space by aligning denoising predictions with target distributions to reestablish severed text-visual associations. However, in multi-concept scenarios, naive reconstruction can cause gradient conflicts and feature entanglement. To address this, we introduce Gradient Field Orthogonalization, which enforces feature orthogonality to prevent mutual interference. Additionally, our Latent Semantic Identification-Guided Sampling (LSIS) ensures stability of the reawakening process via posterior density verification. Extensive experiments demonstrate that LURE enables simultaneous, high-fidelity reawakening of multiple erased concepts across diverse erasure tasks and methods.

The rapid advancements in artificial intelligence have significantly accelerated the adoption of speech recognition technology, leading to its widespread integration across various applications. However, this surge in usage also highlights a critical issue: audio data is highly vulnerable to unauthorized exposure and analysis, posing significant privacy risks for businesses and individuals. This paper introduces an Information-Obfuscation Reversible Adversarial Example (IO-RAE) framework, the pioneering method designed to safeguard audio privacy using reversible adversarial examples. IO-RAE leverages large language models to generate misleading yet contextually coherent content, effectively preventing unauthorized eavesdropping by humans and Automatic Speech Recognition (ASR) systems. Additionally, we propose the Cumulative Signal Attack technique, which mitigates high-frequency noise and enhances attack efficacy by targeting low-frequency signals. Our approach ensures the protection of audio data without degrading its quality or our ability. Experimental evaluations demonstrate the superiority of our method, achieving a targeted misguidance rate of 96.5% and a remarkable 100% untargeted misguidance rate in obfuscating target keywords across multiple ASR models, including a commercial black-box system from Google. Furthermore, the quality of the recovered audio, measured by the Perceptual Evaluation of Speech Quality score, reached 4.45, comparable to high-quality original recordings. Notably, the recovered audio processed by ASR systems exhibited an error rate of 0%, indicating nearly lossless recovery. These results highlight the practical applicability and effectiveness of our IO-RAE framework in protecting sensitive audio privacy.

Autonomous Large Language Model (LLM) agents exhibit significant vulnerability to Indirect Prompt Injection (IPI) attacks. These attacks hijack agent behavior by polluting external information sources, exploiting fundamental trade-offs between security and functionality in existing defense mechanisms. This leads to malicious and unauthorized tool invocations, diverting agents from their original objectives. The success of complex IPIs reveals a deeper systemic fragility: while current defenses demonstrate some effectiveness, most defense architectures are inherently fragmented. Consequently, they fail to provide full integrity assurance across the entire task execution pipeline, forcing unacceptable multi-dimensional compromises among security, functionality, and efficiency. Our method is predicated on a core insight: no matter how subtle an IPI attack, its pursuit of a malicious objective will ultimately manifest as a detectable deviation in the action trajectory, distinct from the expected legitimate plan. Based on this, we propose the Cognitive Control Architecture (CCA), a holistic framework achieving full-lifecycle cognitive supervision. CCA constructs an efficient, dual-layered defense system through two synergistic pillars: (i) proactive and preemptive control-flow and data-flow integrity enforcement via a pre-generated "Intent Graph"; and (ii) an innovative "Tiered Adjudicator" that, upon deviation detection, initiates deep reasoning based on multi-dimensional scoring, specifically designed to counter complex conditional attacks. Experiments on the AgentDojo benchmark substantiate that CCA not only effectively withstands sophisticated attacks that challenge other advanced defense methods but also achieves uncompromised security with notable efficiency and robustness, thereby reconciling the aforementioned multi-dimensional trade-off.

While the deployment of large language models (LLMs) in high-value industries continues to expand, the systematic assessment of their safety against jailbreak and prompt-based attacks remains insufficient. Existing safety evaluation benchmarks and frameworks are often limited by an imbalanced integration of core components (attack, defense, and evaluation methods) and an isolation between flexible evaluation frameworks and standardized benchmarking capabilities. These limitations hinder reliable cross-study comparisons and create unnecessary overhead for comprehensive risk assessment. To address these gaps, we present TeleAI-Safety, a modular and reproducible framework coupled with a systematic benchmark for rigorous LLM safety evaluation. Our framework integrates a broad collection of 19 attack methods (including one self-developed method), 29 defense methods, and 19 evaluation methods (including one self-developed method). With a curated attack corpus of 342 samples spanning 12 distinct risk categories, the TeleAI-Safety benchmark conducts extensive evaluations across 14 target models. The results reveal systematic vulnerabilities and model-specific failure cases, highlighting critical trade-offs between safety and utility, and identifying potential defense patterns for future optimization. In practical scenarios, TeleAI-Safety can be flexibly adjusted with customized attack, defense, and evaluation combinations to meet specific demands. We release our complete code and evaluation results to facilitate reproducible research and establish unified safety baselines.

Diffusion-based image editing has made semantic level image manipulation easy for general users, but it also enables realistic local forgeries that are hard to localize. Existing benchmarks mainly focus on the binary detection of generated images or the localization of manually edited regions and do not reflect the properties of diffusion-based edits, which often blend smoothly into the original content. We present Diffusion-Based Image Editing Area Localization Dataset (DEAL-300K), a large scale dataset for diffusion-based image manipulation localization (DIML) with more than 300,000 annotated images. We build DEAL-300K by using a multi-modal large language model to generate editing instructions, a mask-free diffusion editor to produce manipulated images, and an active-learning change detection pipeline to obtain pixel-level annotations. On top of this dataset, we propose a localization framework that uses a frozen Visual Foundation Model (VFM) together with Multi Frequency Prompt Tuning (MFPT) to capture both semantic and frequency-domain cues of edited regions. Trained on DEAL-300K, our method reaches a pixel-level F1 score of 82.56% on our test split and 80.97% on the external CoCoGlide benchmark, providing strong baselines and a practical foundation for future DIML research.The dataset can be accessed via https://github.com/ymhzyj/DEAL-300K.

Reasoning-augmented Vision-Language Models (RVLMs) rely on safety alignment to prevent harmful behavior, yet their exposed chain-of-thought (CoT) traces introduce new attack surfaces. In this work, we find that the safety alignment of RVLMs can be easily break through a novel attack method termed \textbf{Stealth Fine-Tuning}. Our method elicits harmful reasoning traces through \textbf{segment-level interference} and reuses the self-generated outputs as supervised fine-tuning data. Through a \textbf{turn-based weighted} loss design, yielding a lightweight, distribution-consistent finetuning method. In our experiment, with only 499 samples and under 3 hours on a single A100 (QLoRA), Stealth Fine-Tuning outperforms IDEATOR by 38.52\% ASR while preserving general reasoning ability, as the tuned model retains the original representation distribution. Experiments on AdvBench and several general benchmarks demonstrate that Stealth Fine-Tuning is a low-cost and highly effective way to bypass alignment defenses. \textcolor{red}{\textbf{Disclaimer: This paper contains content that may be disturbing or offensive.}}

Source Code Model learn the proper embeddings from source codes, demonstrating significant success in various software engineering or security tasks. The recent explosive development of LLM extends the family of SCMs,bringing LLMs for code that revolutionize development workflows. Investigating different kinds of SCM vulnerability is the cornerstone for the security and trustworthiness of AI-powered software ecosystems, however, the fundamental one, transferable vulnerability, remains critically underexplored. Existing studies neither offer practical ways, i.e. require access to the downstream classifier of SCMs, to produce effective adversarial samples for adversarial defense, nor give heed to the widely used LLM4Code in modern software development platforms and cloud-based integrated development environments. Therefore, this work systematically studies the intrinsic vulnerability transferability of both traditional SCMs and LLM4Code, and proposes a victim-agnostic approach to generate practical adversarial samples. We design HABITAT, consisting of a tailored perturbation-inserting mechanism and a hierarchical Reinforcement Learning framework that adaptively selects optimal perturbations without requiring any access to the downstream classifier of SCMs. Furthermore, an intrinsic transferability analysis of SCM vulnerabilities is conducted, revealing the potential vulnerability correlation between traditional SCMs and LLM4Code, together with fundamental factors that govern the success rate of victim-agnostic transfer attacks. These findings of SCM vulnerabilities underscore the critical focal points for developing robust defenses in the future. Experimental evaluation demonstrates that our constructed adversarial examples crafted based on traditional SCMs achieve up to 64% success rates against LLM4Code, surpassing the state-of-the-art by over 15%.

Transferable attacks generate adversarial examples on surrogate models to fool unknown victim models, posing real-world threats and growing research interest. Despite focusing on flat losses for transferable adversarial examples, recent studies still fall into suboptimal regions, especially the flat-yet-sharp areas, termed as deceptive flatness. In this paper, we introduce a novel black-box gradient-based transferable attack from a perspective of dual-order information. Specifically, we feasibly propose Adversarial Flatness (AF) to the deceptive flatness problem and a theoretical assurance for adversarial transferability. Based on this, using an efficient approximation of our objective, we instantiate our attack as Adversarial Flatness Attack (AFA), addressing the altered gradient sign issue. Additionally, to further improve the attack ability, we devise MonteCarlo Adversarial Sampling (MCAS) by enhancing the inner-loop sampling efficiency. The comprehensive results on ImageNet-compatible dataset demonstrate superiority over six baselines, generating adversarial examples in flatter regions and boosting transferability across model architectures. When tested on input transformation attacks or the Baidu Cloud API, our method outperforms baselines.

Multimodal large language models (MLLMs) have demonstrated significant utility across diverse real-world applications. But MLLMs remain vulnerable to jailbreaks, where adversarial inputs can collapse their safety constraints and trigger unethical responses. In this work, we investigate jailbreaks in the text-vision multimodal setting and pioneer the observation that visual alignment imposes uneven safety constraints across modalities in MLLMs, thereby giving rise to multimodal safety asymmetry. We then develop PolyJailbreak, a black-box jailbreak method grounded in reinforcement learning. Initially, we probe the model's attention dynamics and latent representation space, assessing how visual inputs reshape cross-modal information flow and diminish the model's ability to separate harmful from benign inputs, thereby exposing exploitable vulnerabilities. On this basis, we systematize them into generalizable and reusable operational rules that constitute a structured library of Atomic Strategy Primitives, which translate harmful intents into jailbreak inputs through step-wise transformations. Guided by the primitives, PolyJailbreak employs a multi-agent optimization process that automatically adapts inputs against the target models. We conduct comprehensive evaluations on a variety of open-source and closed-source MLLMs, demonstrating that PolyJailbreak outperforms state-of-the-art baselines.

With the flourishing prosperity of generative models, manipulated facial images have become increasingly accessible, raising concerns regarding privacy infringement and societal trust. In response, proactive defense strategies embed adversarial perturbations into facial images to counter deepfake manipulation. However, existing methods often face a tradeoff between imperceptibility and defense effectiveness-strong perturbations may disrupt forgeries but degrade visual fidelity. Recent studies have attempted to address this issue by introducing additional visual loss constraints, yet often overlook the underlying gradient conflicts among losses, ultimately weakening defense performance. To bridge the gap, we propose a gradient-projection-based adversarial proactive defense (GRASP) method that effectively counters facial deepfakes while minimizing perceptual degradation. GRASP is the first approach to successfully integrate both structural similarity loss and low-frequency loss to enhance perturbation imperceptibility. By analyzing gradient conflicts between defense effectiveness loss and visual quality losses, GRASP pioneers the design of the gradient-projection mechanism to mitigate these conflicts, enabling balanced optimization that preserves image fidelity without sacrificing defensive performance. Extensive experiments validate the efficacy of GRASP, achieving a PSNR exceeding 40 dB, SSIM of 0.99, and a 100% defense success rate against facial attribute manipulations, significantly outperforming existing approaches in visual quality.

Despite neural ordinary differential equations (Neural ODEs) exhibiting intrinsic robustness under input perturbations due to their dynamical systems nature, recent approaches often involve imposing Lyapunov-based stability conditions to provide formal robustness guarantees. However, a fundamental challenge remains: the tension between robustness and accuracy, primarily stemming from the difficulty in imposing appropriate stability conditions. To address this, we propose an adaptive stable learning framework named Zubov-Net, which innovatively reformulates Zubov's equation into a consistency characterization between regions of attraction (RoAs) and prescribed RoAs (PRoAs). Building on this consistency, we introduce a new paradigm for actively controlling the geometry of RoAs by directly optimizing PRoAs to reconcile accuracy and robustness. Our approach is realized through tripartite losses (consistency, classification, and separation losses) and a parallel boundary sampling algorithm that co-optimizes the Neural ODE and the Lyapunov function. To enhance the discriminativity of Lyapunov functions, we design an input-attention-based convex neural network via a softmax attention mechanism that focuses on equilibrium-relevant features and also serves as weight normalization to maintain training stability in deep architectures. Theoretically, we prove that minimizing the tripartite loss guarantees consistent alignment of PRoAs-RoAs, trajectory stability, and non-overlapping PRoAs. Moreover, we establish stochastic convex separability with tighter probability bounds and fewer dimensionality requirements to justify the convex design in Lyapunov functions. Experimentally, Zubov-Net maintains high classification accuracy while significantly improving robustness against various stochastic noises and adversarial attacks.

With the rapid proliferation of Natural Language Processing (NLP), especially Large Language Models (LLMs), generating adversarial examples to jailbreak LLMs remains a key challenge for understanding model vulnerabilities and improving robustness. In this context, we propose a new black-box attack method that leverages the interpretability of large models. We introduce the Sparse Feature Perturbation Framework (SFPF), a novel approach for adversarial text generation that utilizes sparse autoencoders to identify and manipulate critical features in text. After using the SAE model to reconstruct hidden layer representations, we perform feature clustering on the successfully attacked texts to identify features with higher activations. These highly activated features are then perturbed to generate new adversarial texts. This selective perturbation preserves the malicious intent while amplifying safety signals, thereby increasing their potential to evade existing defenses. Our method enables a new red-teaming strategy that balances adversarial effectiveness with safety alignment. Experimental results demonstrate that adversarial texts generated by SFPF can bypass state-of-the-art defense mechanisms, revealing persistent vulnerabilities in current NLP systems.However, the method's effectiveness varies across prompts and layers, and its generalizability to other architectures and larger models remains to be validated.

Multivariate long-term time series forecasting (MLTSF) models are increasingly deployed in critical domains such as climate, finance, and transportation. Despite their growing importance, the security of MLTSF models against backdoor attacks remains entirely unexplored. To bridge this gap, we propose BadTime, the first effective backdoor attack tailored for MLTSF. BadTime can manipulate hundreds of future predictions toward a target pattern by injecting a subtle trigger. BadTime addresses two key challenges that arise uniquely in MLTSF: (i) the rapid dilution of local triggers over long horizons, and (ii) the extreme sparsity of backdoor signals under stealth constraints. To counter dilution, BadTime leverages inter-variable correlations, temporal lags, and data-driven initialization to design a distributed, lag-aware trigger that ensures effective influence over long-range forecasts. To overcome sparsity, it introduces a hybrid strategy to select valuable poisoned samples and a decoupled backdoor training objective that adaptively adjusts the model's focus on the sparse backdoor signal, ensuring reliable learning at a poisoning rate as low as 1%. Extensive experiments show that BadTime significantly outperforms state-of-the-art (SOTA) backdoor attacks on time series forecasting by extending the attackable horizon from at most 12 timesteps to 720 timesteps (a 60-fold improvement), reducing MAE by over 50% on target variables, and boosting stealthiness by more than 3-fold under anomaly detection.