50,983 papers in this slice of arXiv.
Beining Xu, Hairui Wang, Jiaxin Wang +2
While the privacy risks of multimodal large language models (MLLMs) have drawn significant attention, the unique vulnerabilities of domain-specific MLLMs remain largely underexplored. Focusing on document understanding MLLMs for identity document processing, this paper investigates the privacy issues inherent in Key Information Extraction (KIE) tasks. We reveal that when input images lack sufficient visual evidence, these models often rely on memorized field relations from training data to infer missing content, thereby leaking multiple correlated fields containing sensitive personal information. To mitigate this risk, we make three key contributions.First, we propose the Dynamic Relational Unlearning Framework (DRUF) which comprises a Relational Decoupling Unlearning (RDU) module and a dynamic set update mechanism. It suppresses the leakage of high-risk field pairs while preserving KIE performance.Second, we introduce DocPrivacyBench, a novel benchmark to systematically evaluate a model's susceptibility to privacy leakage under conditions of absent or minimal visual evidence.Third, we evaluate three MLLMs and six unlearning methods using this benchmark, assessing both post-unlearning leakage suppression and utility preservation.Our results demonstrate that existing MLLMs consistently exhibit privacy leakage when visual evidence is scarce, particularly on noisier datasets. In contrast, DRUF outperforms the strongest baseline by improving leakage suppression by 4.8 percentage points, effectively mitigating privacy risks while maintaining robust document information extraction performance.
Denzel Chiuseni, Athanase Bahizire, Silva Hama +1
Smishing detection systems are commonly trained and evaluated on clean, monolingual text. In low-resource settings, however, attackers frequently circumvent these systems through character obfuscation, cross-lingual code-switching, and structural perturbation. This study evaluates adversarial robustness for five model architectures: three classical lexical models (Random Forest, XGBoost, CNN+BiLSTM) and two multilingual transformers (mBERT, XLM-RoBERTa), using a dataset of 27,037 messages. Classical models are subjected to black-box generic attacks, while transformers are evaluated with attention-guided targeting. Each model is tested across three attack types and intensity levels, with performance measured by the Robustness Degradation Ratio (RDR). The results reveal a distinct architectural boundary: classical models experience near-catastrophic failure under character obfuscation and structural perturbation (RDR up to 0.988), whereas transformers demonstrate significantly greater resilience (RDR up to 0.351), with structural perturbation representing their most pronounced vulnerability. Effect-size analysis (Cliff's d) indicates a substantial difference between the two model categories. Within the transformer group, XLM-RoBERTa, despite achieving a higher clean-text baseline, exhibits greater degradation than mBERT. These findings demonstrate that clean-text performance is not a reliable predictor of adversarial robustness. Statistical validation using Mann-Whitney U and Friedman tests confirms that these patterns are attributable to model architecture rather than sampling. The results underscore the necessity for architecture-specific defences and frame smishing detection as an adversarial cybersecurity challenge rather than a static classification task.
Rana Muhammad Ahmed, Sabahat Abbas
Security evaluations of tool-using agents often equate stored labels with behavioral facts. We audit a preserved campaign by tracing 10,200 execution rows to 180 model-bound requests, 45 semantic requests, and 15 observable stimuli. Two schema treatments were delivered, but the planned external payload-family corpus was not. The historical grader exhibited direct treatment leakage: treatment metadata gated the ATTACK_SUCCESS class, so fixed behavior could change class under treatment relabeling. A treatment-blind reconstruction corrects 58 historical ATTACK_SUCCESS or HIJACK_ATTEMPT labels to authorized benign completions while preserving three verified protected-data transfers and one separate unauthorized-forwarding case. The locked v2 census contains exactly zero ATTACK_SUCCESS records, while the forwarding case remains a HIJACK_ATTEMPT at a semantic boundary concerning objective completion. A dual-reviewer blinded concordance review of all 96 requests deemed structurally interpretable by locked v2 produced identical reviewer-consensus classes but differed from the locked codebook on four construct-boundary cases. We contribute a seven-link Integrity Chain and an executable, scope-bounded endpoint-integrity linter. The result is a campaign-bounded measurement audit, not a population attack-rate, model-ranking, defense-efficacy, or causal estimate.
Dorottya Zelenyanszki, Zhe Hou, Kamanashis Biswas +1
Public blockchain data enables large-scale DeFi-related analysis, but many existing approaches are application-specific, difficult to scale, or hard to interpret. This research proposes a scalable, application-agnostic framework for persistent behavioural pattern discovery from large-scale blockchain activity. It constructs behaviour sentences enriched with contract, token and market context, then applies a two-step embedding process: sentence-level embeddings capture individual actions, while sequence-level embeddings capture user behaviour over time. An interpretable behavioural profiler characterizes discovered communities through behavioural motifs, routines, temporal dynamics, entity exposure, and suspiciousness evidence. Evaluation on Ethereum using over 30 million transactions shows that the framework uncovers both routine and malicious behavioural patterns, including decentralised exchange (DEX) trading, NFT activity, phishing, bot operations, oracle manipulation, and rug-pull schemes. Importantly, many patterns remain stable across independent observation windows, enabling the identification of long-term behaviours beyond a single analysis period. The proposed framework combines scalability, interpretability, and persistence analysis, supporting blockchain forensic investigation, behavioural attribution, and threat discovery.
Baihong Chen, Hua Ming, Weifeng Pan +3
Many software systems expose multiple execution modes through command-line options, subcommands, and configuration flags. For such programs, fuzzing depends on both mutated inputs and the invoked mode. Yet evaluations still focus on coverage and bug counts, leaving unclear how execution modes partition, overlap, and miss software structure, and how these differences affect effectiveness. We present an empirical study of software structure under multi-driver fuzzing. We propose a structural abstraction that uses a static call graph as a shared backbone and projects driver-specific dynamic coverage onto it to derive driver-induced subgraphs. Based on this abstraction, we develop a four-phase methodology for backbone construction, fuzzing and profiling, graph-based analysis, and research-question-driven evaluation. We apply it to 27 OSS-Fuzz-derived C/C++ projects, spanning 43 executables and 854 driver configurations. Under the same total budget, multi-driver fuzzing outperforms the best single-driver baseline, increasing covered call-graph nodes by 27.9% and CFG-edge coverage by 73.5%, and revealing 11 unique bugs and abnormal behaviors largely missed by single-driver fuzzing. However, driver contributions are uneven, subgraphs differ substantially in cohesion, fragmentation, modularity, overlap, and residual under-exploration follows recurring regimes rather than a homogeneous tail. These results show that multi-driver fuzzing is fundamentally a structural exploration problem.
Baihong Chen, Tian Xie, Wen Li
Python package security is largely source-centric, yet Python runtimes can execute bytecode directly through .pyc files, compiled-only modules, and marshalled code objects, creating an inspection-execution gap. We present an empirical study of Python bytecode as a security artifact. We measure bytecode exposure in PyPI distributions, evaluate practical analyzability using version-aware tooling, assess CPython runtime robustness under adversarial bytecode, and test source-level reproduction of bytecode findings. Across 1,034,843 collected PyPI artifacts, we identify 7,388 bytecode-containing artifacts, including 228,578 .pyc files and 28,193 artifact-local source-less .pyc files. For modern CPython 3.8-3.14 bytecode, at least one selected decompiler emits source for 204,901 of 204,904 in-scope files, a result measuring emission rather than verified functional equivalence. Tools are non-robust: observed PyPI bytecode triggers managed-code exceptions and timeouts, while adversarial mutated bytecode also drives decompilers into native process failures; together these outcomes yield 17 distinct robustness signatures. Fuzzing produces 1,009 stack-deduplicated runtime findings dominated by pointer-dereference symptoms; 261 groups exhibit potential memory-corruption characteristics, and at least 91.7% of groups reach execution beyond the documented-unsafe ingestion boundary. None reproduce from ordinary Python source. Bytecode is thus a visible ecosystem artifact, a practical analysis target, and a security-relevant interpreter input whose behavior need not match source-level behavior.
Ruofei Qu, Wei Feng, Hongzhan Ma +3
Confidential VMs (CVMs) have become the dominant substrate for sensitive cloud workloads, from financial services to privacy-preserving AI inference. The hardware isolation that protects these CVMs from a malicious cloud also blinds their owners to what runs inside them: kernel rootkits planted via network or supply-chain attacks can hide processes, tamper with kernel data, and exfiltrate model weights under the cover of the same isolation that defends the VM. Tenants therefore need to inspect a running CVM from outside, yet classical VM introspection (VMI) presupposes a trusted Hypervisor, which CVMs exclude from the TCB. The state-of-the-art CVM-VMI system, 00SEVen, restores introspection on AMD SEV-SNP via an in-VM agent at a privileged tier (VMPL0), a mechanism that does not exist on Arm CCA, leaving Realm VMs without any introspection solution. We present RealmEye, the first VMI system for Arm CCA Realm VMs. RealmEye places the entire introspection logic inside the Realm Management Monitor (RMM) at R-EL2, achieving hardware-enforced separation between the monitor and the monitored VM: no agent runs inside the Realm, and the Realm remains unmodified. RealmEye reads Realm memory and registers, suspends the VM for consistent snapshots, and traps page-level accesses, without relying on any in-VM interface. A periodic, self-driven trigger mode keeps scan timing internal to the RMM, preventing the Hypervisor from colluding with in-Realm rootkits. Results are returned to the remote owner over a hardware-attested channel, and a CCA driver backend lets existing tools such as LibVMI and DRAKVUF interoperate with RealmEye unchanged. On the Arm FVP, RealmEye detects process hiding and syscall-table hooking by Diamorphine, and its in-RMM cost is linearly predictable from primitive invocation counts.
Sanjay Kariyappa, Severin Klingler, G. Edward Suh
Tool-using agents consume external data from sources with different levels of trust, yet tool responses rarely identify who produced each component or what it should convey. We show that this gap enables state-corruption attacks, in which attacker-controlled content makes environmental claims beyond the informational authority of its response component and corrupts the agent's perceived environment, making the resulting action appear justified to existing guardrails. We introduce PIPES (Provenance-Informed, Prior-Enforced Screening), which screens response units using semantic priors and source provenance. PIPES uses static field contracts when schemas provide stable expectations, and conditions screening of open-ended content on the pre-response trajectory and trusted provenance metadata. It marks units that violate their semantic prior or the provenance hierarchy; deployments may remove, warn, block, or escalate detected violations. We instantiate atomic removal and evaluate PIPES against adaptive PAIR-style attacks. Across the three VitaBench and three AgentDyn splits with Gemma 4 31B IT as the target agent, PIPES reduces average attack success from 84.7% to 2.3%, while preserving average benign utility (92.5% with PIPES versus 90.6% without defense).
Jesus Salas
Agentic workflows are commonly evaluated by whether they reach the correct outcome. That is insufficient in institutional settings, where a correct action may rely on the wrong authority, an unsupported completion claim, or work made stale by a later change. We define governed execution as work whose decisions, completion, and response to change are supported by inspectable provenance. We present Matrix, a deterministic causal-state layer that records authority and fact dependencies, verifies completion evidence, and selectively invalidates affected work. Across controlled comparisons, governed and direct workflows often reached the same outcomes, but only the governed path consistently preserved governing evidence, refused unsupported closure, and limited recovery to dependent tasks. A role-separated transfer challenge then failed: a deterministically enforced completeness contract severely over-blocked synthetic packets produced outside its authoring context. These results do not establish Matrix as a general accuracy enhancer; they support its primary role as an institutional integrity layer for making agentic work auditable and independently verifiable.
Xiaoyan Feng, Yanjun Zhang, He Zhang +2
Watermarking LLM-generated text is an important task for tracing its provenance. Existing LLM watermarks preserve provenance under editing, but this same robustness allows an adversary to alter critical content while retaining attribution, a vulnerability known as piggyback spoofing. We introduce an innovative watermark that jointly provides provenance and tamper evidence. It co-embeds a robust signal and a fragile signal into each generated token. The signals share the same mechanism but use independent keys and different seeding windows over normalized text, making one resilient to edits and the other sensitive to reader-visible changes. Multiple rounds of unbiased tournament reweighting preserve the expected generation distribution, while a periodic round-allocation pattern controls the trade-off between the two signals. At detection, their scores form a two-dimensional space supporting three decisions: Intact, Tampered, and No-Watermark. Across two large language models and two prompt datasets, our method demonstrates the highest tamper-detection rate among the evaluated methods while maintaining competitive attribution robustness and perplexity. Ablation studies show that reliable three-state detection requires a well-defined notion of intactness, co-embedding of the two signals, and complementary sensitivity to edits.
Saleh Almohaimeed, Saad Almohaimeed, Mousa Jari +2
Retrieval-Augmented Generation (RAG) is widely used to improve the performance of Large Language Models (LLMs) in answering user queries. Existing privacy research on RAG has focused on preventing unauthorized users from accessing sensitive data. However, another important problem that is often overlooked in RAG privacy research is that external generators have access to the query and the retrieved documents, which may contain confidential information that could potentially be misused or accessed for unintended purposes. In this paper, we introduce the Sensitive Entity Alias Generator (SEAG), a privacy-preserving framework that empowers users to utilize powerful third-party generators without disclosing sensitive information. SEAG introduces a lightweight model that locates sensitive entities, generates corresponding aliases, and constructs an entity replacement table. The table is used to replace sensitive words in the user's query and in the retrieved documents before they are forwarded to an external generator. For this purpose, two datasets were constructed: one for fine-tuning SEAG models to generate entity replacement tables, and another for evaluating the entire SEAG framework. The experimental results demonstrate the success of the SEAG framework. As for the User metric, which measures the ability of the model to provide a correct response to the user while hiding sensitive information from the external generator, all SEAG models achieved over 80% accuracy. Additional analysis further evaluated the ability of SEAG models Qwen-3, LLaMA-3.2, and Phi-4 to hide all sensitive entities within given documents. The results show good performance with total accuracies of 77.83%, 76.73%, and 74.91%, respectively.
Shidong Pan, Clark LaChance, Zhen Tao +1
Privacy documents (e.g., privacy policies) are a central mechanism through which digital services disclose data practices and seek user consent. Over the past decades, research on privacy documents has expanded significantly, encompassing not only traditional privacy policies but also short notices (e.g., privacy labels) and interface-level transparency mechanisms. As this research area continues to grow, it has become increasingly difficult to obtain a coherent view of how privacy documents are created, analyzed, evaluated, and maintained across their lifecycle. This SoK provides a unified, lifecycle-oriented view of privacy documents from a software engineering perspective. We systematically review and analyze 290 papers published between 2010 and 2025, organizing them around five research questions that examine how privacy documents are (1) defined and scoped, (2) generated, (3) analyzed and extracted, (4) checked for inconsistencies and noncompliance, and (5) evaluated and improved for usability. Building on our findings, we identify 15 key research trends and 21 open opportunities. We further chart four broader research directions that highlight (i) emerging challenges in AI-centric platforms, (ii) the need for diverse and up-to-date data foundations, (iii) LLM-based unified policy-code analysis, and (iv) dual usability for end-users and developers. We hope this SoK provides a shared foundation for future research on privacy policies and privacy documents.
Junliang Liu, Ruoyu Li, Wenxin Tang +4
LLM agents increasingly rely on third-party skills, using natural-language descriptions for selection and instruction bodies for planning. This progressive-disclosure design exposes two sequential control points to untrusted publishers: a static skill may steer an otherwise correct task onto an unnecessarily costly trajectory. Prior work studies selection manipulation, malicious skill instructions, and tool-chain resource amplification largely separately, leaving their end-to-end composition unclear. We introduce Convergent Detour Hijacking (CDH), a text-only, runtime-independent attack that couples these stages. Under shared semantic cover, a description establishes relevance during selection, while an aligned body reuses that rationale to fabricate plausible dependencies during planning. CDH attracts an attacker-controlled coordinator alongside legitimate skills, recruits unnecessary benign skills into a bounded detour, and then re-enters the original route to preserve task completion. We evaluate it across multiple LLM backends and 491 held-out tasks under single-task and multi-turn conditions. On DeepSeek-V4-Pro, the matched coordinator is selected in 80.02% of tasks; among coordinator-hit runs that complete tasks, token consumption and end-to-end execution time increase by 66.91% and 92.45%, respectively, while aggregate task completion remains comparable. Thus, correct outcomes do not guarantee trajectory integrity or cost safety.
Zhenpeng Li
An unconditional risk bound on automated decisions can be satisfied without automating anything, since a selector that never acts drives the bound to zero. We show this is structural: any risk certificate is defined over a decision contract, the inputs a system acts on plus the semantic relation under which an output counts correct, and weakening either hides base-classifier error. We develop a decision-contract theory: an error-conservation law showing error is only reassigned among harmful automation, human deferral, and semantic masking; a label-free singleton capacity certifying structural incapacity, with a risk-feasible refinement separating recoverable threshold misalignment from risk-constrained incapacity; and a non-degenerate actionability certificate excluding all-abstain solutions by construction. We instantiate this on ATT&CK-aligned alert triage for LLM-based intrusion detection, the setting that exposed the vacuity failure. Across 3 IDS datasets, 6 LLMs, and 4 error-rate thresholds, empirical false-attribution risk stays at or below target in 90.3% of configurations, with 83.4% mean correct automation. The capacity diagnostic explains every low-utility configuration; its refinement separates genuine misalignment from risk-constrained incapacity, confirmed by an exhibited alternative threshold; a training-stability re-run finds no confirmed structural-incapacity instance; and real fine-grained attack-subtype labels confirm the coarsening-transfer identity under a genuine many-to-one map, with small but non-zero masking mass.
Jin Lu, Xuening Han, Yang Zhong +4
Evaluating security vulnerability detection tools requires benchmark datasets with vulnerability-inducing commits (VICs) - the commits that first introduce vulnerabilities into codebases. VICs are essential for determining the full range of vulnerable software versions. Existing vulnerability datasets suffer from limited programming language coverage, restricted patch complexity, and narrow project scope. Through our dual annotation by human experts and an agentic workflow, we create a benchmark - VICBench - of 100 verified VICs for 100 CVEs across 88 projects in Python, Java, and C++, covering 48 CWE types. VICBench features complex real-world vulnerability fixes averaging 38.6 lines and corresponding VICs of 252.5 lines - significantly larger than prior work. Our evaluation shows that state-of-the-art algorithms V-SZZ and LLM4SZZ achieve only 33.3%-40.1% F1, confirming that using existing approaches still entails significant manual effort. VICBench enables robust evaluation of vulnerability detection approaches.
Md Yassir Mottalib, Md Yousuf, Eklachur Rahman Bhuiyan +5
With the increasing complexity of cyber assaults in cloud environments, adaptable security solutions are needed that can support real-time detection and autonomous response. In this paper, we propose a reinforcement learning-based dynamic cyber defense framework. We deploy a Deep Q-Network (DQN) to train effective defensive strategies to counteract the evolving cyberattacks. We leverage the CICIDS2017 dataset for model creation and the UNSW-NB15 dataset for external validation, involving preprocessing of data, feature engineering, and adaptive policy learning. We compare the proposed DQN with decision tree, support vector machine, random forest, XGBoost, and multilayer perceptron models. The proposed DQN achieves an accuracy of 99.72%, a precision of 99.68%, a recall of 99.65%, an F1-score of 99.66%, and an ROC-AUC of 0.999, while the false positive rate is 0.31%, the false negative rate is 0.35%, and the detection latency is 15 ms. The framework achieved 99.54% attack mitigation rate, demonstrating strong adaptive and real-time defensive capabilities. These results demonstrate the potential of reinforcement learning as a powerful and scalable approach for autonomous cybersecurity in modern cloud environments.
Vibha Bhavikatti, Mark Stamp
Recent studies have shown that binary-to-image representations can enable effective machine learning-based results for malware detection and classification. However, performance can vary significantly, depending on the technique used to convert binaries to images. Furthermore, the explainability and interpretability of image-based models is largely unexplored within the malware domain. In this research, we employ Gradient-weighted Class Activation Maps (Grad-CAM) as an eXplainable AI (XAI) tool, which we use to analyze eight distinct image types derived from malware samples. We provide quantitative faithfulness and stability metrics for Grad-CAM heatmaps and we compare these heatmaps to High-Resolution Class Activation Mappings (HiResCAM). We also show that Grad-CAM heatmaps can provide useful information for malware classification. Specifically, we show that a Random Forest model trained on features extracted from Grad-CAM images via a MobileNetV2 Convolutional Neural Network (CNN) model achieves a test accuracy of 0.777 across 17 malware families, exceeding a previous benchmark of 0.750 for this same dataset. A key finding of this research is that for the malware image transformations considered, accuracy and explanation faithfulness do not coincide, e.g., image transformation techniques that produce the most faithful explanations yield only mid-tier accuracy.
Adrián Losada, Hao Qiang Luo-Chen, David Segura +4
The integration of Artificial Intelligence (AI), generally as Machine Learning (ML) algorithms, in all levels and aspects of cellular networks demonstrates the success of data-driven algorithms; for example, the Radio Intelligence Controller (RIC) of the O-RAN paradigm bestows the network with optimised radio resource allocation, load balancing or energy efficiency functions, among others. Nevertheless, this dependency on data opens new security vulnerabilities, as attackers can alter data properties and steer ML models to underperform or degrade. Conversely, the developed mitigation strategies are effective, but they generate a computational load which, in consequence, results in an energy cost generally overlooked, even in the current energy-awareness context. In this work, consumption of a defence technique is characterised, and the challenges raised by the triad of ML accuracy, robustness and energy efficiency are outlined.
Sebastian Garcia, Veronica Valeros, Alya Gomaa +7
Network intrusion detection systems often analyze individual packets or flows, although malicious behavior may develop across many connections and over time. This may limit their ability to combine isolated detections into a coherent assessment of host behavior. Packet-level features may also be too low-level for complex AI-based detection, requiring additional processing to improve accuracy while maintaining a low false-positive rate. We present Slips, a network intrusion detection system that builds host-centered behavioral profiles and organizes activity into time windows. It uses a modular architecture in which independent modules report evidence rather than generating final alerts directly. Slips then accumulates this evidence into host-level decisions. We evaluate Slips against Suricata on an expert-labeled PCAP dataset. At the profile-time-window level, Slips achieved 83% higher recall and a 70% higher F1 score than Suricata, while neither system produced false positives. These results indicate that time-window-based evidence accumulation can produce context-aware decisions that better align with expert judgment.
Jingchuan Ma, Yanhua Liu, Qiaoyun Huang
We classify binary-linear two-term Frobenius-linearized operators L(Y)=AYσ+BY on K3, where K is a finite extension of F2 and σ is a fixed nontrivial Frobenius automorphism of K with fixed field F2. Under a coefficient-rank and binary-kernel condition, if A and B both have K-rank two and L has a one-dimensional kernel over F2, then invertible K-linear input and output changes reduce L, for this fixed σ, to the canonical model (α,β,γ)↦(ασ+α,βσ,γ). The proof constructs the coordinate frames from the two coefficient-kernel directions and the binary kernel. In these coordinates, the first dual output row is exactly the unique nonzero trace-adjoint normal, with an exact K-valued normalization. For pure σ-quadratic almost perfect nonlinear maps, this identifies the orthoderivative by πF(X)TF(X)=1; in odd extension degree it also yields permutation behavior and a bijection from the projective plane to its dual. The triprojective construction of Gologlu and Kolsch and the cubic norm-twist construction of Li, Zhou, Li, and Qu provide two realizations arising from different algebraic constructions. The triprojective case further admits a determinant factorization and a complete dual frame, whereas the norm-twist realization shows that the pure-map consequences do not follow from the operator theorem alone. A natural Gold representation has coefficient-rank pair (3,3), delimiting the rank-two subclass. The normal form also supplies exact extension-field labels for known component-radical and Walsh-support relations.