arXiv:2603.12277v1 Announce Type: cross Abstract: Language models remain vulnerable to prompt injection attacks despite extensive safety training. We trace this failure to role confusion: models infer roles from how text is written, not where it comes from. We design novel role probes to capture how models internally identify "who is speaking." These reveal why prompt injection works: untrusted text that imitates a role inherits that role's authority. We test this insight by injecting spoofed reasoning into user prompts and tool outputs, achieving average success rates of 60% on StrongREJECT and 61% on agent exfiltration, across multiple open- and closed-weight models with near-zero baselines. Strikingly, the degree of internal role confusion strongly predicts attack success before generation begins. Our findings reveal a fundamental gap: security is defined at the interface but authority is assigned in latent space. More broadly, we introduce a unifying, mechanistic framework for prompt injection, demonstrating that diverse prompt-injection attacks exploit the same underlying role-confusion mechanism.

arXiv:2603.12455v1 Announce Type: cross Abstract: The escalating frequency of cyber-attacks poses significant challenges for organisations, particularly small enterprises constrained by limited in-house expertise, insufficient knowledge, and financial resources. This research presents a novel framework that leverages Natural Language Processing to address these challenges through automated mapping of cyber incidents to adversary techniques. We introduce the Cyber Catalog, a knowledge base that systematically integrates CIS Critical Security Controls, MITRE ATT&CK techniques, and SMART metrics. This integrated resource enables organisations to connect threat intelligence directly to actionable controls and measurable outcomes. To operationalise the framework, we fine-tuned all-mpnet-base-v2, a highly regarded sentence-transformers model used to convert text into numerical vectors on an augmented dataset comprising 74,986 incident-technique pairs to enhance semantic similarity between cyber incidents and MITRE ATT&CK techniques. Our fine-tuned model achieved a Spearman correlation of 0.7894 and Pearson correlation of 0.8756, demonstrating substantial improvements over top baseline models including all-mpnet-base-v2, all-distilroberta-v1, and all-MiniLM-L12-v2. Furthermore, our model exhibited significantly lower prediction errors (MAE = 0.135, MSE = 0.027) compared to all baseline models, confirming superior accuracy and consistency. The Cyber Catalog, training dataset, trained model, and implementation code made publicly available to facilitate further research and enable practical deployment in resource-constrained environments. This work bridges the gap between threat intelligence and operational security management, providing an actionable tool for systematic cyber incident response and evidence-based cyber risk management.

arXiv:2603.13028v1 Announce Type: cross Abstract: Diffusion models enable high-fidelity image editing but can also be misused for unauthorized style imitation and harmful content generation. To mitigate these risks, proactive image protection methods embed small, often imperceptible adversarial perturbations into images before sharing to disrupt downstream editing or fine-tuning. However, in realistic post-release scenarios, content owners cannot control downstream processing pipelines, and protections optimized for a surrogate model may fail when attackers use mismatched diffusion pipelines. Existing purification methods can weaken protections but often sacrifice image quality and rarely examine architectural mismatch. We introduce a unified post-release purification framework to evaluate protection survivability under model mismatch. We propose two practical purifiers: VAE-Trans, which corrects protected images via latent-space projection, and EditorClean, which performs instruction-guided reconstruction with a Diffusion Transformer to exploit architectural heterogeneity. Both operate without access to protected images or defense internals. Across 2,100 editing tasks and six representative protection methods, EditorClean consistently restores editability. Compared to protected inputs, it improves PSNR by 3-6 dB and reduces FID by 50-70 percent on downstream edits, while outperforming prior purification baselines by about 2 dB PSNR and 30 percent lower FID. Our results reveal a purify-once, edit-freely failure mode: once purification succeeds, the protective signal is largely removed, enabling unrestricted editing. This highlights the need to evaluate protections under model mismatch and design defenses robust to heterogeneous attackers.

arXiv:2603.13186v1 Announce Type: cross Abstract: Prior approaches for membership privacy preservation usually update or retrain all weights in neural networks, which is costly and can lead to unnecessary utility loss or even more serious misalignment in predictions between training data and non-training data. In this work, we observed three insights: i) privacy vulnerability exists in a very small fraction of weights; ii) however, most of those weights also critically impact utility performance; iii) the importance of weights stems from their locations rather than their values. According to these insights, to preserve privacy, we score critical weights, and instead of discarding those neurons, we rewind only the weights for fine-tuning. We show that, through extensive experiments, this mechanism exhibits outperforming resilience in most cases against Membership Inference Attacks while maintaining utility.

arXiv:2602.16891v2 Announce Type: replace Abstract: Agent development kits (ADKs) provide effective platforms and tooling for constructing agents, and their designs are critical to the constructed agents' performance, especially the functionality for agent topology, tools, and memory. However, current ADKs either lack sufficient functional support or rely on humans to manually design these components, limiting agents' generalizability and overall performance. We propose OpenSage, the first ADK that enables LLMs to automatically create agents with self-generated topology and toolsets while providing comprehensive and structured memory support. OpenSage offers effective functionality for agents to create and manage their own sub-agents and toolkits. It also features a hierarchical, graph-based memory system for efficient management and a specialized toolkit tailored to software engineering tasks. Extensive experiments across three state-of-the-art benchmarks with various backbone models demonstrate the advantages of OpenSage over existing ADKs. We also conduct rigorous ablation studies to demonstrate the effectiveness of our design for each component. We believe OpenSage can pave the way for the next generation of agent development, shifting the focus from human-centered to AI-centered paradigms.

arXiv:2602.23163v2 Announce Type: replace Abstract: Large language models are beginning to show steganographic capabilities. Such capabilities could allow misaligned models to evade oversight mechanisms. Yet principled methods to detect and quantify such behaviours are lacking. Classical definitions of steganography, and detection methods based on them, require a known reference distribution of non-steganographic signals. For the case of steganographic reasoning in LLMs, knowing such a reference distribution is not feasible; this renders these approaches inapplicable. We propose an alternative, \textbf{decision-theoretic view of steganography}. Our central insight is that steganography creates an asymmetry in usable information between agents who can and cannot decode the hidden content (present within a steganographic signal), and this otherwise latent asymmetry can be inferred from the agents' observable actions. To formalise this perspective, we introduce generalised $\mathcal{V}$-information: a utilitarian framework for measuring the amount of usable information within some input. We use this to define the \textbf{steganographic gap} -- a measure that quantifies steganography by comparing the downstream utility of the steganographic signal to agents that can and cannot decode the hidden content. We empirically validate our formalism, and show that it can be used to detect, quantify, and mitigate steganographic reasoning in LLMs.

arXiv:2410.10862v3 Announce Type: replace-cross Abstract: As large language models (LLMs) are overwhelmingly more and more integrated into various applications, ensuring they generate safe responses is a pressing need. Previous studies on alignment have largely focused on general instruction-following but have often overlooked the distinct properties of safety alignment, such as the brittleness of safety mechanisms. To bridge the gap, we propose the Superficial Safety Alignment Hypothesis (SSAH), which posits that safety alignment teaches an otherwise unsafe model to choose the correct reasoning direction-fulfill or refuse users' requests-interpreted as an implicit binary classification task. Through SSAH, we hypothesize that only a few essential components can establish safety guardrails in LLMs. We successfully identify four types of attribute-critical components: Safety Critical Unit (SCU), Utility Critical Unit (UCU), Complex Unit (CU), and Redundant Unit (RU). Our findings show that freezing certain safety-critical components during fine-tuning allows the model to retain its safety attributes while adapting to new tasks. Similarly, we show that leveraging redundant units in the pre-trained model as an "alignment budget" can effectively minimize the alignment tax while achieving the alignment goal. All considered, this paper concludes that the atomic functional unit for safety in LLMs is at the neuron level and underscores that safety alignment should not be complicated. We have code implementation and other information on the project website: https://ssa-h.github.io/.

arXiv:2512.04120v2 Announce Type: replace-cross Abstract: The emergence of open data portals necessitates more attention to protecting sensitive data before datasets get published and exchanged. To do so effectively, we observe the need to refine and broaden our definitions of sensitive data, and argue that the sensitivity of data depends on its context. Following this definition, we introduce a contextual data sensitivity framework building on two core concepts: 1) type contextualization, which considers the type of the data values at hand within the overall context of the dataset or document to assess their true sensitivity, and 2) domain contextualization, which assesses the sensitivity of data values informed by domain-specific information external to the dataset, such as geographic origin of a dataset. Experiments instrumented with language models confirm that: 1) type-contextualization significantly reduces the number of false positives for type-based sensitive data detection and reaches a recall of 94% compared to 63% with commercial tools, and 2) domain-contextualization leveraging sensitivity rule retrieval effectively grounds sensitive data detection in relevant context in non-standard data domains. A case study with humanitarian data experts also illustrates that context-grounded explanations provide useful guidance in manual data auditing processes. We open-source the implementation of the mechanisms and annotated datasets at https://github.com/trl-lab/sensitive-data-detection.

arXiv:2601.18113v3 Announce Type: replace-cross Abstract: LLM-based web agents have become increasingly popular for their utility in daily life and work. However, they exhibit critical vulnerabilities when processing malicious URLs: accepting a disguised malicious URL enables subsequent access to unsafe webpages, which can cause severe damage to service providers and users. Despite this risk, no benchmark currently targets this emerging threat. To address this gap, we propose MalURLBench, the first benchmark for evaluating LLMs' vulnerabilities to malicious URLs. MalURLBench contains 61,845 attack instances spanning 10 real-world scenarios and 7 categories of real malicious websites. Experiments with 12 popular LLMs reveal that existing models struggle to detect elaborately disguised malicious URLs. We further identify and analyze key factors that impact attack success rates and propose URLGuard, a lightweight defense module. We believe this work will provide a foundational resource for advancing the security of web agents. Our code is available at https://github.com/JiangYingEr/MalURLBench.

arXiv:2603.05772v2 Announce Type: replace-cross Abstract: 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.

arXiv:2603.05773v2 Announce Type: replace-cross Abstract: 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.

arXiv:2603.11975v2 Announce Type: replace-cross Abstract: The rapid evolution of embodied agents has accelerated the deployment of household robots in real-world environments. However, unlike structured industrial settings, household spaces introduce unpredictable safety risks, where system limitations such as perception latency and lack of common sense knowledge can lead to dangerous errors. Current safety evaluations, often restricted to static images, text, or general hazards, fail to adequately benchmark dynamic unsafe action detection in these specific contexts. To bridge this gap, we introduce HomeSafe-Bench, a challenging benchmark designed to evaluate Vision-Language Models (VLMs) on unsafe action detection in household scenarios. HomeSafe-Bench is contrusted via a hybrid pipeline combining physical simulation with advanced video generation and features 438 diverse cases across six functional areas with fine-grained multidimensional annotations. Beyond benchmarking, we propose Hierarchical Dual-Brain Guard for Household Safety (HD-Guard), a hierarchical streaming architecture for real-time safety monitoring. HD-Guard coordinates a lightweight FastBrain for continuous high-frequency screening with an asynchronous large-scale SlowBrain for deep multimodal reasoning, effectively balancing inference efficiency with detection accuracy. Evaluations demonstrate that HD-Guard achieves a superior trade-off between latency and performance, while our analysis identifies critical bottlenecks in current VLM-based safety detection.

arXiv:2603.07379v1 Announce Type: new Abstract: Retrieval-Augmented Generation (RAG) systems are increasingly evolving into agentic architectures where large language models autonomously coordinate multi-step reasoning, dynamic memory management, and iterative retrieval strategies. Despite rapid industrial adoption, current research lacks a systematic understanding of Agentic RAG as a sequential decision-making system, leading to highly fragmented architectures, inconsistent evaluation methodologies, and unresolved reliability risks. This Systematization of Knowledge (SoK) paper provides the first unified framework for understanding these autonomous systems. We formalize agentic retrieval-generation loops as finite-horizon partially observable Markov decision processes, explicitly modeling their control policies and state transitions. Building upon this formalization, we develop a comprehensive taxonomy and modular architectural decomposition that categorizes systems by their planning mechanisms, retrieval orchestration, memory paradigms, and tool-invocation behaviors. We further analyze the critical limitations of traditional static evaluation practices and identify severe systemic risks inherent to autonomous loops, including compounding hallucination propagation, memory poisoning, retrieval misalignment, and cascading tool-execution vulnerabilities. Finally, we outline key doctoral-scale research directions spanning stable adaptive retrieval, cost-aware orchestration, formal trajectory evaluation, and oversight mechanisms, providing a definitive roadmap for building reliable, controllable, and scalable agentic retrieval systems.

arXiv:2603.07427v1 Announce Type: new Abstract: As Large Language Models (LLMs) evolve into autonomous agents, existing safety evaluations face a fundamental trade-off: manual benchmarks are costly, while LLM-based simulators are scalable but suffer from logic hallucination. We present AutoControl Arena, an automated framework for frontier AI risk evaluation built on the principle of logic-narrative decoupling. By grounding deterministic state in executable code while delegating generative dynamics to LLMs, we mitigate hallucination while maintaining flexibility. This principle, instantiated through a three-agent framework, achieves over 98% end-to-end success and 60% human preference over existing simulators. To elicit latent risks, we vary environmental Stress and Temptation across X-Bench (70 scenarios, 7 risk categories). Evaluating 9 frontier models reveals: (1) Alignment Illusion: risk rates surge from 21.7% to 54.5% under pressure, with capable models showing disproportionately larger increases; (2) Scenario-Specific Safety Scaling: advanced reasoning improves robustness for direct harms but worsens it for gaming scenarios; and (3) Divergent Misalignment Patterns: weaker models cause non-malicious harm while stronger models develop strategic concealment.

arXiv:2603.06862v1 Announce Type: cross Abstract: Artifact Evaluation (AE) is essential for ensuring the transparency and reliability of research, closing the gap between exploratory work and real-world deployment is particularly important in cybersecurity, particularly in IoT and CPSs, where large-scale, heterogeneous, and privacy-sensitive data meet safety-critical actuation. Yet, manual reproducibility checks are time-consuming and do not scale with growing submission volumes. In this work, we demonstrate that Large Language Models (LLMs) can provide powerful support for AE tasks: (i) text-based reproducibility rating, (ii) autonomous sandboxed execution environment preparation, and (iii) assessment of methodological pitfalls. Our reproducibility-assessment toolkit yields an accuracy of over 72% and autonomously sets up execution environments for 28% of runnable cybersecurity artifacts. Our automated pitfall assessment detects seven prevalent pitfalls with high accuracy ($F_1$ > 92%). Hence, the toolkit significantly reduces reviewer effort and, when integrated into established AE processes, could incentivize authors to submit higher-quality and more reproducible artifacts. IoT, CPS, and cybersecurity conferences and workshops may integrate the toolkit into their peer-review processes to support reviewers' decisions on awarding artifact badges, improving the overall sustainability of the process.

arXiv:2603.07090v1 Announce Type: cross Abstract: As Joint Audio-Visual Generation Models see widespread commercial deployment, embedding watermarks has become essential for protecting vendor copyright and ensuring content provenance. However, existing techniques suffer from an architectural mismatch by treating modalities as decoupled entities, exposing a critical Binding Vulnerability. Adversaries exploit this via Swap Attacks by replacing authentic audio with malicious deepfakes while retaining the watermarked video. Because current detectors rely on independent verification ($Video_{wm}\vee Audio_{wm}$), they incorrectly authenticate the manipulated content, falsely attributing harmful media to the original vendor and severely damaging their reputation. To address this, we propose mAVE (Manifold Audio-Visual Entanglement), the first watermarking framework natively designed for joint architectures. mAVE cryptographically binds audio and video latents at initialization without fine-tuning, defining a Legitimate Entanglement Manifold via Inverse Transform Sampling. Experiments on state-of-the-art models (LTX-2, MOVA) demonstrate that mAVE guarantees performance-losslessness and provides an exponential security bound against Swap Attacks. Achieving near-perfect binding integrity ($>99\%$), mAVE offers a robust cryptographic defense for vendor copyright.

arXiv:2603.07116v1 Announce Type: cross Abstract: As autonomous AI agents increasingly populate the Internet, a novel security challenge arises: "Is this entity an AI agent?" It is a new entity-type verification problem with no established solution. We formalize the problem through a three-class entity taxonomy (Human, Script, Agent) based on a verifiable agentic capability vector (action, reasoning, and memory). A timing threshold t exploits the asymmetric hardness between human cognition and AI processing to separate the three classes. We define the Agentic Capability Verification Problem (ACVP) through three necessity primitives, each testing one capability dimension. Building on this foundation, we introduce aCAPTCHA (Agent CAPTCHA), a time-constrained security game for agent admission whose security rests on ACVP hardness under t. We instantiate aCAPTCHA through time-bounded natural-language understanding as a multi-round HTTP verification protocol, and evaluate it with preliminary agent trials that validate the protocol's soundness and completeness. aCAPTCHA provides a composable, infrastructure-free admission gate for any service where entity-type verification is required.

arXiv:2603.07191v1 Announce Type: cross Abstract: Autonomous agents powered by large language models introduce a class of execution-layer vulnerabilities -- prompt injection, retrieval poisoning, and uncontrolled tool invocation -- that existing guardrails fail to address systematically. In this work, we propose the Layered Governance Architecture (LGA), a four-layer framework comprising execution sandboxing (L1), intent verification (L2), zero-trust inter-agent authorization (L3), and immutable audit logging (L4). To evaluate LGA, we construct a bilingual benchmark (Chinese original, English via machine translation) of 1,081 tool-call samples -- covering prompt injection, RAG poisoning, and malicious skill plugins -- and apply it to OpenClaw, a representative open-source agent framework. Experimental results on Layer 2 intent verification with four local LLM judges (Qwen3.5-4B, Llama-3.1-8B, Qwen3.5-9B, Qwen2.5-14B) and one cloud judge (GPT-4o-mini) show that all five LLM judges intercept 93.0-98.5% of TC1/TC2 malicious tool calls, while lightweight NLI baselines remain below 10%. TC3 (malicious skill plugins) proves harder at 75-94% IR among judges with meaningful precision-recall balance, motivating complementary enforcement at Layers 1 and 3. Qwen2.5-14B achieves the best local balance (98% IR, approximately 10-20% FPR); a two-stage cascade (Qwen3.5-9B->GPT-4o-mini) achieves 91.9-92.6% IR with 1.9-6.7% FPR; a fully local cascade (Qwen3.5-9B->Qwen2.5-14B) achieves 94.7-95.6% IR with 6.0-9.7% FPR for data-sovereign deployments. An end-to-end pipeline evaluation (n=100) demonstrates that all four layers operate in concert with 96% IR and a total P50 latency of approximately 980 ms, of which the non-judge layers contribute only approximately 18 ms. Generalization to the external InjecAgent benchmark yields 99-100% interception, confirming robustness beyond our synthetic data.

arXiv:2603.07452v1 Announce Type: cross Abstract: Backdoor mechanisms have traditionally been studied as security threats that compromise the integrity of machine learning models. However, the same mechanism -- the conditional activation of specific behaviors through input triggers -- can also serve as a controllable and auditable interface for trustworthy model behavior. In this work, we present \textbf{Backdoor4Good (B4G)}, a unified benchmark and framework for \textit{beneficial backdoor} applications in large language models (LLMs). Unlike conventional backdoor studies focused on attacks and defenses, B4G repurposes backdoor conditioning for Beneficial Tasks that enhance safety, controllability, and accountability. It formalizes beneficial backdoor learning under a triplet formulation $(T, A, U)$, representing the \emph{Trigger}, \emph{Activation mechanism}, and \emph{Utility function}, and implements a benchmark covering four trust-centric applications. Through extensive experiments across Llama3.1-8B, Gemma-2-9B, Qwen2.5-7B, and Llama2-13B, we show that beneficial backdoors can achieve high controllability, tamper-resistance, and stealthiness while preserving clean-task performance. Our findings demonstrate new insights that backdoors need not be inherently malicious; when properly designed, they can serve as modular, interpretable, and beneficial building blocks for trustworthy AI systems. Our code and datasets are available at https://github.com/bboylyg/BackdoorLLM/B4G.

arXiv:2603.07460v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly integrated into safety-critical workflows, yet existing security analyses remain fragmented and often isolate model behavior from the broader system context. This work introduces a goal-driven risk assessment framework for LLM-powered systems that combines system modeling with Attack-Defense Trees (ADTrees) and Common Vulnerability Scoring System (CVSS)-based exploitability scoring to support structured, comparable analysis. We demonstrate the framework through a healthcare case study, modeling multi-step attack paths targeting intervention in medical procedures, leakage of electronic health record (EHR) data, and disruption of service availability. Our analysis indicates that threats spanning (i) conventional cyber, (ii) adversarial ML, and (iii) conversational attacks that manipulate prompts or context often consolidate into a small number of dominant paths and shared system choke points, enabling targeted defenses to yield meaningful reductions in path exploitability. By systematically comparing defense portfolios, we align these risks with established vulnerability management practices and provide a domain-agnostic workflow applicable to other LLM-enabled critical systems.