arXiv subjects

Shiqian Zhao

Publications and source records attributed to Shiqian Zhao.

At least 19 recordsLinked to original sources

arXiv2026

ECLIPSE: Self-Evolving Stealthy Prompt Injection Attack against Long-Horizon Agentic Systems

Recently, large language model (LLM) agents, such as Codex, Claude Code, and OpenClaw, have become capable of planning and executing long-horizon tasks through repeated tool calls. This capability also creates new opportunities for prompt injection. Existing attacks either place the malicious objective in one explicit instruction, making it easy to detect, or distribute the intent across multiple execution stages, making successful completion unreliable. In this work, we propose ECLIPSE, a self-evolving and stealthy prompt-injection framework for long-horizon agentic systems. ECLIPSE combines direct user-prompt injection with indirect tool-side injection through two components. On the one hand, Stealthy Attack Trajectory Synthesis uses a sandbox to generate and iteratively verify candidate tool chains, then renders a verified chain as a natural one-shot prompt to serve as the direct instruction. Then, Tool-Chain Steering transfers this plan to the target environment through Static Workflow Encoding (SWE), which embeds state-transition cues in target-tool descriptions, and Dynamic Trajectory Correction (DTC), which supplies corrective signals when execution deviates from the planned chain. To enable systematic evaluation, we further introduce LASE-Bench, a long-horizon agent-safety benchmark with 120 malicious tasks and 198 unique tools; 96.7% of its tasks make at least five tool calls. The experimental results show that ECLIPSE is highly effective: it achieves up to 96.7% attack success without defense and 69.2% under the common safety filter, exceeding the strongest baseline by 27.5% in the defended setting. Evaluations against representative defenses further show that existing safeguards do not reliably defend it, which raises the need for more effective defenses.

cs.CR↗

arXiv2026

When Claws Remember but Do Not Tell: Stealthy Memory Injection in Persistent Personal Agents

Persistent personal agents combine long-term memory with access to users' external environments, enabling personalized foreground assistance and proactive background execution. This integration also creates a new path to compromise: untrusted external content can be silently written into persistent memory and later reused as trusted state. We study this threat as stealth memory injection, in which a remote black-box adversary delivers a single email payload that must induce the agent to write poisoned memory, stay hidden in the agent's response to the user, and affect future behavior. We introduce WhisperBench, a 108-case benchmark spanning five risk categories and both fact and preference poisoning. Built on a real IMAP/SMTP workflow and an authentic email agent skill, it enables full-cycle evaluation of stealth memory injection attacks. To enable this black-box attack under single-email delivery and without runtime feedback, we propose MemGhost, a one-shot payload generation framework. MemGhost uses an environment proxy to emulate persistent-agent execution and an objective proxy to convert memory adoption and conversational stealth into dense rubric-based rewards, then trains the attacker policy with supervised fine-tuning and reinforcement learning. Across 56 held-out test cases, MemGhost achieves 87.5% end-to-end success on OpenClaw with GPT-5.4 and 71.4% on Claude Code SDK with Sonnet 4.6. It also transfers across personal-agent architectures (NanoClaw and Hermes Agent) and memory backends (filesystem and vector-based Mem0), and remains effective against input-level, model-level, and system-level defenses. These results suggest that persistent memory can turn ordinary external processing into a practical pathway for long-term agent compromise.

cs.CR↗

arXiv2026

No Strong Evidence for Plasma Lensing in FRB 20240114A

FRB~20240114A is an extremely active repeating fast radio burst for which plasma lensing has been proposed to explain its burst-rate variations, spectral evolution, and apparently ``carbon-copy'' burst pairs. Using FAST data and publicly available Parkes observations, we test this interpretation with a one-dimensional Gaussian plasma-lens model. Although the burst-rate enhancements can be fitted separately, the corresponding magnification peaks and demagnification troughs are offset by far more than predicted and show no consistent periodicity. Moreover, with more than 10,000 bursts detected, a few apparently ``carbon-copy'' pairs can readily occur by chance. The burst bandwidth is not systematically narrower during the proposed lensing interval, nor are the burst energies significantly enhanced during the predicted magnification interval. These results provide no compelling evidence that a single Gaussian plasma lens explains the observed variability, which is more likely dominated by intrinsic source activity.

astro-ph.HE↗

arXiv2026

Random Polarization Position Angle Behaviors across Bursts of Repeating Fast Radio Bursts

Fast radio bursts (FRBs), highly polarized, mostly have a nearly constant polarization position angle (PA) during each burst. Their PAs are observed to vary from burst to burst, with the statistical properties remaining stable across different observation sessions. We found that the intrinsic PAs of repeating FRBs are approximately Gaussian distributed, suggesting that the emission likely originates from a localized region within the neutron star's magnetosphere. A periodicity search of the PA time series using the Lomb-Scargle periodogram reveals no credible periodic signal in the period range from 10 ms to $10^7$ ms, and similar analyses of several active observations also yield null detections. We interpret these properties by extending the rotating vector model to include a dynamically evolving magnetosphere, in which the effective magnetic axis varies from burst to burst due to stochastic perturbations. In this framework, the observed PA distributions can naturally arise from geometric projection effects, and the absence of periodicity reflects the random wandering of the magnetic axis within a confined region. This scenario provides a natural explanation for both repeating and apparently non-repeating FRBs.

astro-ph.HE↗

arXiv2026

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

We identify a critical security vulnerability in mainstream Claw personal AI agents: untrusted content encountered during heartbeat-driven background execution can silently pollute agent memory and subsequently influence user-facing behavior without the user's awareness. This vulnerability arises from an architectural design shared across the Claw ecosystem: heartbeat background execution runs in the same session as user-facing conversation, so content ingested from any external source monitored in the background (including email, message channels, news feeds, code repositories, and social platforms) can enter the same memory context used for foreground interaction, often with limited user visibility and without clear source provenance. We formalize this process as an Exposure (E) $\rightarrow$ Memory (M) $\rightarrow$ Behavior (B) pathway: misinformation encountered during heartbeat execution enters the agent's short-term session context, potentially gets written into long-term memory, and later shapes downstream user-facing behavior. We instantiate this pathway in an agent-native social setting using MissClaw, a controlled research replica of Moltbook. We find that (1) social credibility cues, especially perceived consensus, are the dominant driver of short-term behavioral influence, with misleading rates up to 61%; (2) routine memory-saving behavior can promote short-term pollution into durable long-term memory at rates up to 91%, with cross-session behavioral influence reaching 76%; (3) under naturalistic browsing with content dilution and context pruning, pollution still crosses session boundaries. Overall, prompt injection is not required: ordinary social misinformation is sufficient to silently shape agent memory and behavior under heartbeat-driven background execution.

cs.CR↗

arXiv2026

Depth Charge: Jailbreak Large Language Models from Deep Safety Attention Heads

Currently, open-sourced large language models (OSLLMs) have demonstrated remarkable generative performance. However, as their structure and weights are made public, they are exposed to jailbreak attacks even after alignment. Existing attacks operate primarily at shallow levels, such as the prompt or embedding level, and often fail to expose vulnerabilities rooted in deeper model components, which creates a false sense of security for successful defense. In this paper, we propose \textbf{\underline{S}}afety \textbf{\underline{A}}ttention \textbf{\underline{H}}ead \textbf{\underline{A}}ttack (\textbf{SAHA}), an attention-head-level jailbreak framework that explores the vulnerability in deeper but insufficiently aligned attention heads. SAHA contains two novel designs. Firstly, we reveal that deeper attention layers introduce more vulnerability against jailbreak attacks. Based on this finding, \textbf{SAHA} introduces \textit{Ablation-Impact Ranking} head selection strategy to effectively locate the most vital layer for unsafe output. Secondly, we introduce a boundary-aware perturbation method, \textit{i.e. Layer-Wise Perturbation}, to probe the generation of unsafe content with minimal perturbation to the attention. This constrained perturbation guarantees higher semantic relevance with the target intent while ensuring evasion. Extensive experiments show the superiority of our method: SAHA improves ASR by 14\% over SOTA baselines, revealing the vulnerability of the attack surface on the attention head. Our code is available at https://anonymous.4open.science/r/SAHA.

cs.CR↗

arXiv2026

Knowing without Acting: The Disentangled Geometry of Safety Mechanisms in Large Language Models

Safety alignment is often conceptualized as a monolithic process wherein harmfulness detection automatically triggers refusal. However, the persistence of jailbreak attacks suggests a fundamental mechanistic decoupling. We propose the \textbf{\underline{D}}isentangled \textbf{\underline{S}}afety \textbf{\underline{H}}ypothesis \textbf{(DSH)}, positing that safety computation operates on two distinct subspaces: a \textit{Recognition Axis} ($\mathbf{v}_H$, ``Knowing'') and an \textit{Execution Axis} ($\mathbf{v}_R$, ``Acting''). Our geometric analysis reveals a universal ``Reflex-to-Dissociation'' evolution, where these signals transition from antagonistic entanglement in early layers to structural independence in deep layers. To validate this, we introduce \textit{Double-Difference Extraction} and \textit{Adaptive Causal Steering}. Using our curated \textsc{AmbiguityBench}, we demonstrate a causal double dissociation, effectively creating a state of ``Knowing without Acting.'' Crucially, we leverage this disentanglement to propose the \textbf{Refusal Erasure Attack (REA)}, which achieves State-of-the-Art attack success rates by surgically lobotomizing the refusal mechanism. Furthermore, we uncover a critical architectural divergence, contrasting the \textit{Explicit Semantic Control} of Llama3.1 with the \textit{Latent Distributed Control} of Qwen2.5. The code and dataset are available at https://anonymous.4open.science/r/DSH.

cs.CR↗

arXiv2026

When Memory Becomes a Vulnerability: Towards Multi-turn Jailbreak Attacks against Text-to-Image Generation Systems

Modern text-to-image (T2I) generation systems (e.g., DALL$\cdot$E 3) exploit the memory mechanism, which captures key information in multi-turn interactions for faithful generation. Despite its practicality, the security analyses of this mechanism have fallen far behind. In this paper, we reveal that it can exacerbate the risk of jailbreak attacks. Previous attacks fuse the unsafe target prompt into one ultimate adversarial prompt, which can be easily detected or lead to the generation of non-unsafe images due to under- or over-detoxification. In contrast, we propose embedding the malice at the inception of the chat session in memory, addressing the above limitations. Specifically, we propose Inception, the first multi-turn jailbreak attack against real-world text-to-image generation systems that explicitly exploits their memory mechanisms. Inception is composed of two key modules: segmentation and recursion. We introduce Segmentation, a semantic-preserving method that generates multi-round prompts. By leveraging NLP analysis techniques, we design policies to decompose a prompt, together with its malicious intent, according to sentence structure, thereby evading safety filters. Recursion further addresses the challenge posed by unsafe sub-prompts that cannot be separated through simple segmentation. It firstly expands the sub-prompt, then invokes segmentation recursively. To facilitate multi-turn adversarial prompts crafting, we build VisionFlow, an emulation T2I system that integrates two-stage safety filters and industrial-grade memory mechanisms. The experiment results show that Inception successfully allures unsafe image generation, surpassing the SOTA by a 20.0\% margin in attack success rate. We also conduct experiments on the real-world commercial T2I generation platforms, further validating the threats of Inception in practice.

cs.CV↗

arXiv2026

Inference-time Alignment via Sparse Junction Steering

Token-level steering has emerged as a pivotal approach for inference-time alignment, enabling fine grained control over large language models by modulating their output distributions without parameter updates. While effective, existing methods rely on dense intervention at every decoding step. This persistent manipulation not only incurs substantial computational overhead but also risks compromising generation quality by excessively drifting from the model's intrinsic distribution. In this work, we show that dense intervention is unnecessary and propose Sparse Inference time Alignment (SIA), which performs sparse junction steering by intervening only at critical decision points along the generation trajectory. Our key insight is that high entropy junctions mark pivotal decision points in the generation trajectory and are particularly susceptible to misalignment, indicating the need to introduce alignment related reward signals at these points. Extensive experiments across different model families and alignment objectives show that steering only 20% to 80% of tokens achieves superior alignment-efficiency trade offs. For strong base models such as Qwen3, intervening on as few as 20% of tokens matches or even surpasses heavily post-trained instruct models. This sparsity enables stronger guidance while better preserving the model's native distribution, integrates seamlessly with search based methods such as Best-of-N, and reduces computational cost by up to 6x.

cs.CL↗

arXiv2026

WADBERT: Dual-channel Web Attack Detection Based on BERT Models

Web attack detection is the first line of defense for securing web applications, designed to preemptively identify malicious activities. Deep learning-based approaches are increasingly popular for their advantages: automatically learning complex patterns and extracting semantic features from HTTP requests to achieve superior detection performance. However, existing methods are less effective in embedding irregular HTTP requests, even failing to model unordered parameters and achieve attack traceability. In this paper, we propose an effective web attack detection model, named WADBERT. It achieves high detection accuracy while enabling the precise identification of malicious parameters. To this end, we first employ Hybrid Granularity Embedding (HGE) to generate fine-grained embeddings for URL and payload parameters. Then, URLBERT and SecBERT are respectively utilized to extract their semantic features. Further, parameter-level features (extracted by SecBERT) are fused through a multi-head attention mechanism, resulting in a comprehensive payload feature. Finally, by feeding the concatenated URL and payload features into a linear classifier, a final detection result is obtained. The experimental results on CSIC2010 and SR-BH2020 datasets validate the efficacy of WADBERT, which respectively achieves F1-scores of 99.63% and 99.50%, and significantly outperforms state-of-the-art methods.

cs.CR↗

arXiv2026

Towards Effective Prompt Stealing Attack against Text-to-Image Diffusion Models

Text-to-Image (T2I) models, represented by DALL$\cdot$E and Midjourney, have gained huge popularity for creating realistic images. The quality of these images relies on the carefully engineered prompts, which have become valuable intellectual property. While skilled prompters showcase their AI-generated art on markets to attract buyers, this business incidentally exposes them to \textit{prompt stealing attacks}. Existing state-of-the-art attack techniques reconstruct the prompts from a fixed set of modifiers (i.e., style descriptions) with model-specific training, which exhibit restricted adaptability and effectiveness to diverse showcases (i.e., target images) and diffusion models. To alleviate these limitations, we propose Prometheus, a training-free, proxy-in-the-loop, search-based prompt-stealing attack, which reverse-engineers the valuable prompts of the showcases by interacting with a local proxy model. It consists of three innovative designs. First, we introduce dynamic modifiers, as a supplement to static modifiers used in prior works. These dynamic modifiers provide more details specific to the showcases, and we exploit NLP analysis to generate them on the fly. Second, we design a contextual matching algorithm to sort both dynamic and static modifiers. This offline process helps reduce the search space of the subsequent step. Third, we interact with a local proxy model to invert the prompts with a greedy search algorithm. Based on the feedback guidance, we refine the prompt to achieve higher fidelity. The evaluation results show that Prometheus successfully extracts prompts from popular platforms like PromptBase and AIFrog against diverse victim models, including Midjourney, Leonardo.ai, and DALL$\cdot$E, with an ASR improvement of 25.0\%. We also validate that Prometheus is resistant to extensive potential defenses, further highlighting its severity in practice.

cs.CR↗

arXiv2025

Towards DM-free search for Fast Radio Bursts with Machine Learning -- I. An implementation on multibeam data

Searching for fleeting radio transients like fast radio bursts (FRBs) with wide-field radio telescopes has become a common challenge in data-intensive science. Conventional algorithms normally cost enormous time to seek candidates by finding the correct dispersion measures, of which the process is so-called dedispersion. Here we present a novel scheme to identify FRB signals from raw data without dedispersion using Machine Learning (ML). Under the data environment for multibeam receivers, we train the EfficientNet model and achieve both exceeding 92% accuracy and precision in FRB recognition. We find that the searching efficiency can be significantly enhanced without the procedure of dedispersion compared with conventional softwares like TransientX and presto. Specifically, the impact of radio frequency interference (RFI) for single-beam and multibeam data has been investigated, and we find ML can naturally mitigate RFI under the multibeam environment. Finally, we validate the trained model on actual data from the current FRB surveys carried out by the Five-hundred-meter Aperture Spherical radio Telescope, which provides considerable potential for real implementation in the future.

astro-ph.IM↗

arXiv2025

Mask Image Watermarking

We present MaskWM, a simple, efficient, and flexible framework for image watermarking. MaskWM has two variants: (1) MaskWM-D, which supports global watermark embedding, watermark localization, and local watermark extraction for applications such as tamper detection; (2) MaskWM-ED, which focuses on local watermark embedding and extraction, offering enhanced robustness in small regions to support fine-grined image protection. MaskWM-D builds on the classical encoder-distortion layer-decoder training paradigm. In MaskWM-D, we introduce a simple masking mechanism during the decoding stage that enables both global and local watermark extraction. During training, the decoder is guided by various types of masks applied to watermarked images before extraction, helping it learn to localize watermarks and extract them from the corresponding local areas. MaskWM-ED extends this design by incorporating the mask into the encoding stage as well, guiding the encoder to embed the watermark in designated local regions, which improves robustness under regional attacks. Extensive experiments show that MaskWM achieves state-of-the-art performance in global and local watermark extraction, watermark localization, and multi-watermark embedding. It outperforms all existing baselines, including the recent leading model WAM for local watermarking, while preserving high visual quality of the watermarked images. In addition, MaskWM is highly efficient and adaptable. It requires only 20 hours of training on a single A6000 GPU, achieving 15x computational efficiency compared to WAM. By simply adjusting the distortion layer, MaskWM can be quickly fine-tuned to meet varying robustness requirements.

cs.CV↗

arXiv2025

P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs

During fine-tuning, large language models (LLMs) are increasingly vulnerable to data-poisoning backdoor attacks, which compromise their reliability and trustworthiness. However, existing defense strategies suffer from limited generalization: they only work on specific attack types or task settings. In this study, we propose Poison-to-Poison (P2P), a general and effective backdoor defense algorithm. P2P injects benign triggers with safe alternative labels into a subset of training samples and fine-tunes the model on this re-poisoned dataset by leveraging prompt-based learning. This enforces the model to associate trigger-induced representations with safe outputs, thereby overriding the effects of original malicious triggers. Thanks to this robust and generalizable trigger-based fine-tuning, P2P is effective across task settings and attack types. Theoretically and empirically, we show that P2P can neutralize malicious backdoors while preserving task performance. We conduct extensive experiments on classification, mathematical reasoning, and summary generation tasks, involving multiple state-of-the-art LLMs. The results demonstrate that our P2P algorithm significantly reduces the attack success rate compared with baseline models. We hope that the P2P can serve as a guideline for defending against backdoor attacks and foster the development of a secure and trustworthy LLM community.

cs.CR↗

arXiv2025

T2V-OptJail: Discrete Prompt Optimization for Text-to-Video Jailbreak Attacks

In recent years, fueled by the rapid advancement of diffusion models, text-to-video (T2V) generation models have achieved remarkable progress, with notable examples including Pika, Luma, Kling, and Open-Sora. Although these models exhibit impressive generative capabilities, they also expose significant security risks due to their vulnerability to jailbreak attacks, where the models are manipulated to produce unsafe content such as pornography, violence, or discrimination. Existing works such as T2VSafetyBench provide preliminary benchmarks for safety evaluation, but lack systematic methods for thoroughly exploring model vulnerabilities. To address this gap, we are the first to formalize the T2V jailbreak attack as a discrete optimization problem and propose a joint objective-based optimization framework, called T2V-OptJail. This framework consists of two key optimization goals: bypassing the built-in safety filtering mechanisms to increase the attack success rate, preserving semantic consistency between the adversarial prompt and the unsafe input prompt, as well as between the generated video and the unsafe input prompt, to enhance content controllability. In addition, we introduce an iterative optimization strategy guided by prompt variants, where multiple semantically equivalent candidates are generated in each round, and their scores are aggregated to robustly guide the search toward optimal adversarial prompts. We conduct large-scale experiments on several T2V models, covering both open-source models and real commercial closed-source models. The experimental results show that the proposed method improves 11.4% and 10.0% over the existing state-of-the-art method in terms of attack success rate assessed by GPT-4, attack success rate assessed by human accessors, respectively, verifying the significant advantages of the method in terms of attack effectiveness and content control.

cs.CV↗

arXiv2025

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment

The remarkable success of Large Language Models (LLMs) has illuminated a promising pathway toward achieving Artificial General Intelligence for both academic and industrial communities, owing to their unprecedented performance across various applications. As LLMs continue to gain prominence in both research and commercial domains, their security and safety implications have become a growing concern, not only for researchers and corporations but also for every nation. Currently, existing surveys on LLM safety primarily focus on specific stages of the LLM lifecycle, e.g., deployment phase or fine-tuning phase, lacking a comprehensive understanding of the entire "lifechain" of LLMs. To address this gap, this paper introduces, for the first time, the concept of "full-stack" safety to systematically consider safety issues throughout the entire process of LLM training, deployment, and eventual commercialization. Compared to the off-the-shelf LLM safety surveys, our work demonstrates several distinctive advantages: (I) Comprehensive Perspective. We define the complete LLM lifecycle as encompassing data preparation, pre-training, post-training, deployment and final commercialization. To our knowledge, this represents the first safety survey to encompass the entire lifecycle of LLMs. (II) Extensive Literature Support. Our research is grounded in an exhaustive review of over 800+ papers, ensuring comprehensive coverage and systematic organization of security issues within a more holistic understanding. (III) Unique Insights. Through systematic literature analysis, we have developed reliable roadmaps and perspectives for each chapter. Our work identifies promising research directions, including safety in data generation, alignment techniques, model editing, and LLM-based agent systems. These insights provide valuable guidance for researchers pursuing future work in this field.

cs.CR↗

arXiv2025

Three Minds, One Legend: Jailbreak Large Reasoning Model with Adaptive Stacked Ciphers

Recently, Large Reasoning Models (LRMs) have demonstrated superior logical capabilities compared to traditional Large Language Models (LLMs), gaining significant attention. Despite their impressive performance, the potential for stronger reasoning abilities to introduce more severe security vulnerabilities remains largely underexplored. Existing jailbreak methods often struggle to balance effectiveness with robustness against adaptive safety mechanisms. In this work, we propose SEAL, a novel jailbreak attack that targets LRMs through an adaptive encryption pipeline designed to override their reasoning processes and evade potential adaptive alignment. Specifically, SEAL introduces a stacked encryption approach that combines multiple ciphers to overwhelm the models reasoning capabilities, effectively bypassing built-in safety mechanisms. To further prevent LRMs from developing countermeasures, we incorporate two dynamic strategies - random and adaptive - that adjust the cipher length, order, and combination. Extensive experiments on real-world reasoning models, including DeepSeek-R1, Claude Sonnet, and OpenAI GPT-o4, validate the effectiveness of our approach. Notably, SEAL achieves an attack success rate of 80.8% on GPT o4-mini, outperforming state-of-the-art baselines by a significant margin of 27.2%. Warning: This paper contains examples of inappropriate, offensive, and harmful content.

cs.CL↗

arXiv2025

Search for Fast Radio Bursts and radio pulsars from pulsing Ultraluminous X-ray Sources

We conducted targeted fast radio burst (FRB) and pulsar searches on eight pulsing ultraluminous X-ray sources (PULXs) using the Five-hundred-meter Aperture Spherical Radio Telescope (FAST) and the Parkes 64-meter Radio Telescope (Murriyang) to investigate whether PULXs could be progenitors of FRBs. FAST carried out 12 observations of four PULXs, totaling 8 hours, while Parkes conducted 12 observations of the remaining four PULXs, totaling 11 hours. No significant signals were detected through single-pulse and periodic searches, covering a dispersion measure (DM) range of 0-5000 pc cm$^{-3}$, placing stringent upper limits on the radio flux density from these sources. The results imply that accretion processes and dense stellar winds in PULXs likely suppress or attenuate potential coherent emission in radio band. Additionally, the beaming factor and luminosity of FRBs associated with PULXs, as well as the highly relativistic and magnetized nature of their outflows, may limit detectability. Non-detection yielded from the observations covering the full orbital phases of PULXs can also constrain the theoretical models that link FRB emission to highly magnetized neutron stars in binary systems.

astro-ph.HE↗