Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Cryptography and Security

cs.CR Watch Cryptography and Security

Monday 30 March

Cryptography and Security · Cross-listed from cs.LG

FL-PBM: Pre-Training Backdoor Mitigation for Federated Learning

Osama Wehbi, Sarhad Arisdakessian, Omar Abdel Wahab et al. (+3 more)

Backdoor attacks pose a significant threat to the integrity and reliability of Artificial Intelligence (AI) models, enabling adversaries to manipulate model behavior by injecting poisoned data with hidden triggers. These attacks can lead to severe consequences, especially in critical applications such as autonomous driving, healthcare, and finance. Detecting and mitigating backdoor attacks is crucial across the lifespan of model's phases, including pre-training, in-training, and post-training. In this paper, we propose Pre-Training Backdoor Mitigation for Federated Learning (FL-PBM), a novel defense mechanism that proactively filters poisoned data on the client side before model training in a federated learning (FL) environment. The approach consists of three stages: (1) inserting a benign trigger into the data to establish a controlled baseline, (2) applying Principal Component Analysis (PCA) to extract discriminative features and assess the separability of the data, (3) performing Gaussian Mixture Model (GMM) clustering to identify potentially malicious data samples based on their distribution in the PCA-transformed space, and (4) applying a targeted blurring technique to disrupt potential backdoor triggers. Together, these steps ensure that suspicious data is detected early and sanitized effectively, thereby minimizing the influence of backdoor triggers on the global model. Experimental evaluations on image-based datasets demonstrate that FL-PBM reduces attack success rates by up to 95% compared to baseline federated learning (FedAvg) and by 30 to 80% relative to state-of-the-art defenses (RDFL and LPSF). At the same time, it maintains over 90% clean model accuracy in most experiments, achieving better mitigation without degrading model performance.
Read the full abstract

All authors Osama Wehbi, Sarhad Arisdakessian, Omar Abdel Wahab, Azzam Mourad, Hadi Otrok, Jamal Bentahar

Cryptography and Security

Safeguarding LLMs Against Misuse and AI-Driven Malware Using Steganographic Canaries

Md Raz, Venkata Sai Charan Putrevu, Meet Udeshi et al. (+3 more)

AI-powered malware increasingly exploits cloud-hosted generative-AI services and large language models (LLMs) as analysis engines for reconnaissance and code generation. Simultaneously, enterprise uploads expose sensitive documents to third-party AI vendors. Both threats converge at the AI service ingestion boundary, yet existing defenses focus on endpoints and network perimeters, leaving organizations with limited visibility once plaintext reaches an LLM service. To address this, we present a framework based on steganographic canary files: realistic documents carrying cryptographically derived identifiers embedded via complementary encoding channels. A pre-ingestion filter extracts and verifies these identifiers before LLM processing, enabling passive, format-agnostic detection without semantic classification. We support two modes of operation where Mode A marks existing sensitive documents with layered symbolic encodings (whitespace substitution, zero-width character insertion, homoglyph substitution), while Mode B generates synthetic canary documents using linguistic steganography (arithmetic coding over GPT-2), augmented with compatible symbolic layers. We model increasing document pre-processing and adversarial capability for both modes via a four-tier transport-transform taxonomy: All methods achieve 100% identifier recovery under benign and sanitization workflows (Tiers 1-2). The hybrid Mode B maintains 97% through targeted adversarial transforms (Tier 3). An end-to-end case study against an LLM-orchestrated ransomware pipeline confirms that both modes detect and block canary-bearing uploads before file encryption begins. To our knowledge, this is the first framework to systematically combine symbolic and linguistic text steganography into layered canary documents for detecting unauthorized LLM processing, evaluated against a transport-threat taxonomy tailored to AI malware.
Read the full abstract

All authors Md Raz, Venkata Sai Charan Putrevu, Meet Udeshi, Prashanth Krishnamurthy, Farshad Khorrami, Ramesh Karri

Cryptography and Security

Interpretable Ensemble Learning for Network Traffic Anomaly Detection: A SHAP-based Explainable AI Framework for Embedded Systems Security

Wanru Shao

Network security threats in embedded systems pose significant challenges to critical infrastructure protection. This paper presents a comprehensive framework combining ensemble learning methods with explainable artificial intelligence (XAI) techniques for robust anomaly detection in network traffic. We evaluate multiple machine learning models including Random Forest, Gradient Boosting, Support Vector Machines, and ensemble methods on a real-world network traffic dataset containing 19 features derived from packet-level and frequency domain characteristics. Our experimental results demonstrate that ensemble methods achieve superior performance, with Random Forest attaining 90% accuracy and an AUC of 0.617 on validation data. Furthermore, we employ SHAP (SHapley Additive exPlanations) analysis to provide interpretable insights into model predictions, revealing that packet_count_5s,inter_arrival_time, and spectral_entropy are the most influential features for anomaly detection. The integration of XAI techniques enhances model trustworthiness and facilitates deployment in security-critical embedded systems where interpretability is paramount.
Read the full abstract

Cryptography and Security · Cross-listed from cs.LG

Mitigating Backdoor Attacks in Federated Learning Using PPA and MiniMax Game Theory

Osama Wehbi, Sarhad Arisdakessian, Omar Abdel Wahab et al. (+3 more)

Federated Learning (FL) is witnessing wider adoption due to its ability to benefit from large amounts of scattered data while preserving privacy. However, despite its advantages, federated learning suffers from several setbacks that directly impact the accuracy, and the integrity of the global model it produces. One of these setbacks is the presence of malicious clients who actively try to harm the global model by injecting backdoor data into their local models while trying to evade detection. The objective of such clients is to trick the global model into making false predictions during inference, thereby compromising the integrity and trustworthiness of the global model on which honest stakeholders rely. To mitigate such mischievous behavior, we propose FedBBA (Federated Backdoor and Behavior Analysis). The proposed model aims to dampen the effect of such clients on the final accuracy, creating more resilient federated learning environments. We engineer our approach through the combination of (1) a reputation system to evaluate and track client behavior, (2) an incentive mechanism to reward honest participation and penalize malicious behavior, and (3) game theoretical models with projection pursuit analysis (PPA) to dynamically identify and minimize the impact of malicious clients on the global model. Extensive simulations on the German Traffic Sign Recognition Benchmark (GTSRB) and Belgium Traffic Sign Classification (BTSC) datasets demonstrate that FedBBA reduces the backdoor attack success rate to approximately 1.1%–11% across various attack scenarios, significantly outperforming state-of-the-art defenses like RDFL and RoPE, which yielded attack success rates between 23% and 76%, while maintaining high normal task accuracy (~95%–98%).
Read the full abstract

All authors Osama Wehbi, Sarhad Arisdakessian, Omar Abdel Wahab, Anderson Avila, Azzam Mourad, Hadi Otrok

Cryptography and Security

Empowering Mobile Networks Security Resilience by using Post-Quantum Cryptography

Ricardo Alves Faval, Rodrigo Moreira and Flávio de Oliveira Silva

The transition to a cloud-native 5G Service-Based Architecture (SBA) improves scalability but exposes control-plane signaling to emerging quantum threats, including Harvest-Now, Decrypt-Later (HNDL) attacks. While NIST has standardized post-quantum cryptography (PQC), practical, deployable integration in operational 5G cores remains underexplored. This work experimentally integrates NIST-standardized ML-KEM-768 and ML-DSA into an open-source 5G core (free5GC) using a sidecar proxy pattern that preserves unmodified network functions (NFs). Implemented on free5GC, we compare three deployments: (i) native HTTPS/TLS, (ii) TLS sidecar, and (iii) PQC-enabled sidecar. Measurements at the HTTP/2 request-response boundary over repeated independent runs show that PQC increases end-to-end Service-Based Interface (SBI) latency to approximately 54 ms, adding a deterministic 48-49 ms overhead relative to the classical baseline, while maintaining tightly bounded variance (IQR <= 0.2 ms, CV < 0.4%). We also quantify the impact of Certification Authority (CA) security levels, identifying certificate validation as a tunable contributor to overall delay. Overall, the results demonstrate that sidecar-based PQC insertion enables a non-disruptive and operationally predictable migration path for quantum-resilient 5G signaling.
Read the full abstract

Cryptography and Security · Cross-listed from cs.CV

TGIF2: Extended Text-Guided Inpainting Forgery Dataset & Benchmark

Hannes Mareen, Dimitrios Karageorgiou, Paschalis Giakoumoglou et al. (+3 more)

Generative AI has made text-guided inpainting a powerful image editing tool, but at the same time a growing challenge for media forensics. Existing benchmarks, including our text-guided inpainting forgery (TGIF) dataset, show that image forgery localization (IFL) methods can localize manipulations in spliced images but struggle not in fully regenerated (FR) images, while synthetic image detection (SID) methods can detect fully regenerated images but cannot perform localization. With new generative inpainting models emerging and the open problem of localization in FR images remaining, updated datasets and benchmarks are needed. We introduce TGIF2, an extended version of TGIF, that captures recent advances in text-guided inpainting and enables a deeper analysis of forensic robustness. TGIF2 augments the original dataset with edits generated by FLUX.1 models, as well as with random non-semantic masks. Using the TGIF2 dataset, we conduct a forensic evaluation spanning IFL and SID, including fine-tuning IFL methods on FR images and generative super-resolution attacks. Our experiments show that both IFL and SID methods degrade on FLUX.1 manipulations, highlighting limited generalization. Additionally, while fine-tuning improves localization on FR images, evaluation with random non-semantic masks reveals object bias. Furthermore, generative super-resolution significantly weakens forensic traces, demonstrating that common image enhancement operations can undermine current forensic pipelines. In summary, TGIF2 provides an updated dataset and benchmark, which enables new insights into the challenges posed by modern inpainting and AI-based image enhancements. TGIF2 is available at https://github.com/IDLabMedia/tgif-dataset.
Read the full abstract

All authors Hannes Mareen, Dimitrios Karageorgiou, Paschalis Giakoumoglou, Peter Lambert, Symeon Papadopoulos, Glenn Van Wallendael

Cryptography and Security · Cross-listed from cs.CV

Detection of Adversarial Attacks in Robotic Perception

Ziad Sharawy, Mohammad Nakshbandi and Sorin Mihai Grigorescu

Deep Neural Networks (DNNs) achieve strong performance in semantic segmentation for robotic perception but remain vulnerable to adversarial attacks, threatening safety-critical applications. While robustness has been studied for image classification, semantic segmentation in robotic contexts requires specialized architectures and detection strategies.
Read the full abstract

Cryptography and Security

"What Did It Actually Do?": Understanding Risk Awareness and Traceability for Computer-Use Agents

Zifan Peng and Mingchen Li

Personalized computer-use agents are rapidly moving from expert communities into mainstream use. Unlike conventional chatbots, these systems can install skills, invoke tools, access private resources, and modify local environments on users' behalf. Yet users often do not know what authority they have delegated, what the agent actually did during task execution, or whether the system has been safely removed afterward. We investigate this gap as a combined problem of risk understanding and post-hoc auditability, using OpenClaw as a motivating case. We first build a multi-source corpus of the OpenClaw ecosystem, including incidents, advisories, malicious-skill reports, news coverage, tutorials, and social-media narratives. We then conduct an interview study to examine how users and practitioners understand skills, autonomy, privilege, persistence, and uninstallation. Our findings suggest that participants often recognized these systems as risky in the abstract, but lacked concrete mental models of what skills can do, what resources agents can access, and what changes may remain after execution or removal. Motivated by these findings, we propose AgentTrace, a traceability framework and prototype interface for visualizing agent actions, touched resources, permission history, provenance, and persistent side effects. A scenario-based evaluation suggests that traceability-oriented interfaces can improve understanding of agent behavior, support anomaly detection, and foster more calibrated trust.
Read the full abstract

Cryptography and Security · Cross-listed from cs.NI

Shy Guys: A Light-Weight Approach to Detecting Robots on Websites

Rémi Van Boxem, Tom Barbette, Cristel Pelsser et al. (+1 more)

Automated bots now account for roughly half of all web requests, and an increasing number deliberately spoof their identity to either evade detection or to not respect robots.txt. Existing countermeasures are either resource-intensive (JavaScript challenges, CAPTCHAs), cost-prohibitive (commercial solutions), or degrade the user experience. This paper proposes a lightweight, passive approach to bot detection that combines user-agent string analysis with favicon-based heuristics, operating entirely on standard web server logs with no client-side interaction. We evaluate the method on over 4.6 million requests containing 54,945 unique user-agent strings collected from website hosted all around the earth. Our approach detects 67.7% of bot traffic while maintaining a false-positive rate of 3%, outperforming state of the art (less than 20%). This method can serve as a first line of defence, routing only genuinely ambiguous requests to active challenges and preserving the experience of legitimate users.
Read the full abstract

All authors Rémi Van Boxem, Tom Barbette, Cristel Pelsser, Ramin Sadre

Cryptography and Security

Democratizing Federated Learning with Blockchain and Multi-Task Peer Prediction

Leon Witt, Kentaroh Toyoda, Wojciech Samek et al. (+1 more)

The synergy between Federated Learning and blockchain has been considered promising; however, the computationally intensive nature of contribution measurement conflicts with the strict computation and storage limits of blockchain systems. We propose a novel concept to decentralize the AI training process using blockchain technology and Multi-task Peer Prediction. By leveraging smart contracts and cryptocurrencies to incentivize contributions to the training process, we aim to harness the mutual benefits of AI and blockchain. We discuss the advantages and limitations of our design.
Read the full abstract

All authors Leon Witt, Kentaroh Toyoda, Wojciech Samek, Dan Li

Cryptography and Security · Cross-listed from cs.LG

Label-efficient Training Updates for Malware Detection over Time

Luca Minnei, Cristian Manca, Giorgio Piras et al. (+6 more)

Machine Learning (ML)-based detectors are becoming essential to counter the proliferation of malware. However, common ML algorithms are not designed to cope with the dynamic nature of real-world settings, where both legitimate and malicious software evolve. This distribution drift causes models trained under static assumptions to degrade over time unless they are continuously updated. Regularly retraining these models, however, is expensive, since labeling new acquired data requires costly manual analysis by security experts. To reduce labeling costs and address distribution drift in malware detection, prior work explored active learning (AL) and semi-supervised learning (SSL) techniques. Yet, existing studies (i) are tightly coupled to specific detector architectures and restricted to a specific malware domain, resulting in non-uniform comparisons; and (ii) lack a consistent methodology for analyzing the distribution drift, despite the critical sensitivity of the malware domain to temporal changes. In this work, we bridge this gap by proposing a model-agnostic framework that evaluates an extensive set of AL and SSL techniques, isolated and combined, for Android and Windows malware detection. We show that these techniques, when combined, can reduce manual annotation costs by up to 90% across both domains while achieving comparable detection performance to full-labeling retraining. We also introduce a methodology for feature-level drift analysis that measures feature stability over time, showing its correlation with the detector performance. Overall, our study provides a detailed understanding of how AL and SSL behave under distribution drift and how they can be successfully combined, offering practical insights for the design of effective detectors over time.
Read the full abstract

All authors Luca Minnei, Cristian Manca, Giorgio Piras, Angelo Sotgiu, Maura Pintor, Daniele Ghiani, Davide Maiorca, Giorgio Giacinto, Battista Biggio

Cryptography and Security

Cryptanalysis of a Lightweight RFID Authentication Protocol Based on a Variable Matrix Encryption Algorithm

Hongjun Wu

Recently, a two-way RFID authentication protocol based on the AM-SUEO-DBLTKM variable matrix encryption algorithm was proposed for low-cost mobile RFID systems. Its design combines adaptive modulus selection, self-updating matrix ordering, and transpose/block-based matrix generation. In this paper, we show that the protocol has structural weaknesses. First, the underlying primitive remains a linear transformation modulo a session modulus, with no nonlinear confusion layer and no ciphertext chaining. Second, in the lightweight setting emphasized by the original paper, the update space is very small: there are only a few modulus choices, only four matrix-order choices when two secret matrices are used, and only a limited family of DBLTKM-generated matrices. Third, the correctness requirements of the protocol impose nontrivial constraints on the sizes of the modulus and plaintext coordinates, weakening the claimed entropy of the secret quantities. Building on these observations, we describe a multi-session algebraic attack path. Under repeated reuse of the same matrix and modulus – an event plausible because of the small update space – ciphertexts corresponding to $N_t$, $N_t+1$, $N_r$, and $N_r+1$ reveal a full column of the matrix. Across sessions, transpose-based matrix generation helps recover additional entries of the secret matrices, while the remaining entries can be obtained later from ordinary ciphertext equations. We then show that candidate factors of the session moduli can be tested by solving reduced equations for secret $S$ across many sessions and checking for mutually consistent solutions. This, in turn, enables recovery of candidate 64-bit moduli and the remaining protocol secrets. Taken together, our results indicate that the protocol is structurally insecure and admits a realistic route to full compromise in the lightweight parameter regime advocated for deployment.
Read the full abstract

Cryptography and Security

VulnScout-C: A Lightweight Transformer for C Code Vulnerability Detection

Aymen Lassoued, Nacef Mbarek, Bechir Dardouri et al. (+3 more)

Vulnerability detection in C programs is a critical challenge in software security. Although large language models (LLMs) achieve strong detection performance, their multi-billion-parameter scale makes them impractical for integration into development workflows requiring low latency and continuous analysis. We introduce VULNSCOUT-C, a compact transformer architecture with 693M total parameters (353M active during inference), derived from the Qwen model family and optimized for C code vulnerability detection. Alongside the model, we present VULNSCOUT, a new 33,565-sample curated dataset generated through a controlled multi-agent pipeline with formal verification, designed to fill coverage gaps in existing benchmarks across underrepresented CWE categories. Evaluated on a standardized C vulnerability detection benchmark, VULNSCOUT-C outperforms all evaluated baselines, including state-of-the-art reasoning LLMs and commercial static analysis tools, while offering a fraction of their inference cost. These results demonstrate that task-specialized compact architectures can match or even outperform the detection capability of models orders of magnitude larger, making continuous, low-latency vulnerability analysis practical within real-world development workflows.
Read the full abstract

All authors Aymen Lassoued, Nacef Mbarek, Bechir Dardouri, Bassem Ouni, Qing Li, Fakhri Karray

Cryptography and Security · Cross-listed from cs.IT

Secret Key Rate Analysis of RIS-Assisted THz MIMO CV-QKD Systems under Access-Constrained Eavesdropping

Sushil Kumar, Soumya P. Dash and George C. Alexandropoulos

A reconfigurable intelligent surface (RIS)-assisted multiple-input multiple-output (MIMO) continuous-variable quantum key distribution (CV-QKD) system operating at terahertz (THz) frequencies, in which a transmitter, Alice, encodes secret keys using Gaussian-modulated coherent states and communicates them to a legitimate receiver, Bob, is considered in this paper. The composite wireless channel, comprising the direct Alice-Bob link and the RIS-assisted reflected link, is modeled as a passive linear Gaussian quantum channel, enabling a unitary dilation that preserves the canonical commutation relations. The security of the considered system is investigated under collective Gaussian entangling attacks by introducing a practical access-constrained eavesdropping model, in which an eavesdropper, Eve, is assumed to access only the environmental modes associated with physically accessible propagation segments. A unified analytical framework is developed to derive the achievable secret key rate (SKR) across all single-segment, pairwise-segment, and full-segment access scenarios, assuming homodyne detection and reverse reconciliation at Bob. Furthermore, an optimization framework is developed to determine the optimal RIS phase configuration matrix and transmitter/receiver beam-splitter parameters that maximize the SKR performance. The resulting optimization problem is efficiently solved using particle swarm optimization. Numerical results are presented to demonstrate the system's performance with respect to various free parameters. It is showcased that the secrecy performance strongly depends on Eve's accessible propagation segments, with the Alice-RIS channel constituting the most security-critical segment.
Read the full abstract

Cryptography and Security

Evaluating Privilege Usage of Agents with Real-World Tools

Quan Zhang, Lianhang Fu, Lvsi Lian et al. (+5 more)

Equipping LLM agents with real-world tools can substantially improve productivity. However, granting agents autonomy over tool use also transfers the associated privileges to both the agent and the underlying LLM. Improper privilege usage may lead to serious consequences, including information leakage and infrastructure damage. While several benchmarks have been built to study agents' security, they often rely on pre-coded tools and restricted interaction patterns. Such crafted environments differ substantially from the real-world, making it hard to assess agents' security capabilities in critical privilege control and usage. Therefore, we propose GrantBox, a security evaluation sandbox for analyzing agent privilege usage. GrantBox automatically integrates real-world tools and allows LLM agents to invoke genuine privileges, enabling the evaluation of privilege usage under prompt injection attacks. Our results indicate that while LLMs exhibit basic security awareness and can block some direct attacks, they remain vulnerable to more sophisticated attacks, resulting in an average attack success rate of 84.80% in carefully crafted scenarios.
Read the full abstract

All authors Quan Zhang, Lianhang Fu, Lvsi Lian, Gwihwan Go, Yujue Wang, Chijin Zhou, Yu Jiang, Geguang Pu

Cryptography and Security

Silent Guardians: Independent and Secure Decision Tree Evaluation Without Chatter

Jinyuan Li and Liang Feng Zhang

As machine learning as a service (MLaaS) gains increasing popularity, it raises two critical challenges: privacy and verifiability. For privacy, clients are reluctant to disclose sensitive private information to access MLaaS, while model providers must safeguard their proprietary models. For verifiability, clients lack reliable mechanisms to ensure that cloud servers execute model inference correctly. Decision trees are widely adopted in MLaaS due to their popularity, interpretability, and broad applicability in domains like medicine and finance. In this context, outsourcing decision tree evaluation (ODTE) enables both clients and model providers to offload their sensitive data and decision tree models to the cloud securely. However, existing ODTE schemes often fail to address both privacy and verifiability simultaneously. To bridge this gap, we propose $\sf PVODTE$, a novel two-server private and verifiable ODTE protocol that leverages homomorphic secret sharing and a MAC-based verification mechanism. $\sf PVODTE$ eliminates the need for server-to-server communication, enabling independent computation by each cloud server. This “non-interactive” setting addresses the latency and synchronization bottlenecks of prior arts, making it uniquely suitable for wide-area network (WAN) deployments. To our knowledge, $\sf PVODTE$ is the first two-server ODTE protocol that eliminates server-to-server communication. Furthermore, $\sf PVODTE$ achieves security against malicious servers, where servers cannot learn anything about the client's input or the providers' decision tree models, and servers cannot alter the inference result without being detected.
Read the full abstract

Cryptography and Security · Cross-listed from cs.LG

ORACAL: A Robust and Explainable Multimodal Framework for Smart Contract Vulnerability Detection with Causal Graph Enrichment

Tran Duong Minh Dai, Triet Huynh Minh Le, M. Ali Babar et al. (+2 more)

Although Graph Neural Networks (GNNs) have shown promise for smart contract vulnerability detection, they still face significant limitations. Homogeneous graph models fail to capture the interplay between control flow and data dependencies, while heterogeneous graph approaches often lack deep semantic understanding, leaving them susceptible to adversarial attacks. Moreover, most black-box models fail to provide explainable evidence, hindering trust in professional audits. To address these challenges, we propose ORACAL (Observable RAG-enhanced Analysis with CausAL reasoning), a heterogeneous multimodal graph learning framework that integrates Control Flow Graph (CFG), Data Flow Graph (DFG), and Call Graph (CG). ORACAL selectively enriches critical subgraphs with expert-level security context from Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs), and employs a causal attention mechanism to disentangle true vulnerability indicators from spurious correlations. For transparency, the framework adopts PGExplainer to generate subgraph-level explanations identifying vulnerability triggering paths. Experiments on large-scale datasets demonstrate that ORACAL achieves state-of-the-art performance, outperforming MANDO-HGT, MTVHunter, GNN-SC, and SCVHunter by up to 39.6 percentage points, with a peak Macro F1 of 91.28% on the primary benchmark. ORACAL maintains strong generalization on out-of-distribution datasets with 91.8% on CGT Weakness and 77.1% on DAppScan. In explainability evaluation, PGExplainer achieves 32.51% Mean Intersection over Union (MIoU) against manually annotated vulnerability triggering paths. Under adversarial attacks, ORACAL limits performance degradation to approximately 2.35% F1 decrease with an Attack Success Rate (ASR) of only 3%, surpassing SCVHunter and MANDO-HGT which exhibit ASRs ranging from 10.91% to 18.73%.
Read the full abstract

All authors Tran Duong Minh Dai, Triet Huynh Minh Le, M. Ali Babar, Van-Hau Pham, Phan The Duy

Cryptography and Security

Seeing the Unseen: Rethinking Illicit Promotion Detection with In-Context Learning

Sangyi Wu, Junpu Guo and Xianghang Mi

Illicit online promotion is a persistent threat that evolves to evade detection. Existing moderation systems remain tethered to platform-specific supervision and static taxonomies, a reactive paradigm that struggles to generalize across domains or uncover novel threats. This paper presents a systematic study of In-Context Learning (ICL) as a unified framework for illicit promotion detection. Through rigorous analysis, we show that properly configured ICL achieves performance comparable to fine-tuned models using 22x fewer labeled examples. We demonstrate three key capabilities: (1) Generalization to unseen threats: ICL generalizes to new illicit categories without category-specific demonstrations, with a performance drop of less than 6% for most evaluated categories. (2) Autonomous discovery: A novel two-stage pipeline distills 2,900 free-form labels into coherent taxonomies, surfacing eight previously undocumented illicit categories such as usury and illegal immigration. (3) Cross-platform generalization: Deployed on 200,000 real-world samples from search engines and Twitter without adaptation, ICL achieves 92.6% accuracy. Furthermore, 61.8% of its uniquely flagged samples correspond to borderline or obfuscated content missed by existing detectors. Our findings position ICL as a new paradigm for content moderation, combining the precision of specialized classifiers with cross-platform generalization and autonomous threat discovery. By shifting to inference-time reasoning, ICL offers a path toward proactively adaptive moderation systems.
Read the full abstract

Cryptography and Security

Kill-Chain Canaries: Stage-Level Tracking of Prompt Injection Across Attack Surfaces and Five Production LLMs

Haochuan Kevin Wang and Zechen Zhang

Multi-agent LLM systems now read documents, web pages and tool results on behalf of users, yet their resistance to prompt injection is usually reported as one number: did the attack succeed? We introduce a kill-chain canary method that plants a unique token in every injected payload and records the furthest of four stages it reaches (Exposed -> Persisted -> Relayed -> Executed), across 950 runs, five production LLMs, six attack surfaces, and five defense conditions. Exposure was 100% among runs that called the tool; the outcomes differ downstream. Claude Haiku 4.5 and Claude Sonnet 4.5 executed none of their 164 text-surface attacks, and in the text relay the canary token never appeared in a memory write (0/40); GPT-4o-mini executed 53% of its attacks. Four findings follow. (1) A Claude writer kept the canary token out of shared memory in every relay run we report; one cross-model pairing (Claude writer, GPT-4o-mini reader, n = 3) is consistent with this protecting the reader, and other pairings were not tested. (2) As readers, the Claude models executed 0/40 raw pre-seeded injections, but Claude Haiku 4.5 executed 2/3 injections relayed by GPT-4o-mini; whether relayed injections are harder to refuse than raw ones is an open question. (3) DeepSeek Chat went from 0/24 on pre-seeded memory to 8/8 on tool results, scenarios that also differ in task and payload format; white-text PDF payloads, invisible on the rendered page, succeeded at least as often as visible ones. (4) pi_detector and write_filter failed on channels they do not inspect, spotlighting failed on content it wraps, and write_filter blocked the PDF relay but not the text relay, a difference we cannot explain. Code and run logs are publicly released: https://github.com/KevinChunye/prompt_injection
Read the full abstract

Cryptography and Security

FedFG: Privacy-Preserving and Robust Federated Learning via Flow-Matching Generation

Ruiyang Wang, Rong Pan and Zhengan Yao

Federated learning (FL) enables distributed clients to collaboratively train a global model using local private data. Nevertheless, recent studies show that conventional FL algorithms still exhibit deficiencies in privacy protection, and the server lacks a reliable and stable aggregation rule for updating the global model. This situation creates opportunities for adversaries: on the one hand, they may eavesdrop on uploaded gradients or model parameters, potentially leaking benign clients' private data; on the other hand, they may compromise clients to launch poisoning attacks that corrupt the global model. To balance accuracy and security, we propose FedFG, a robust FL framework based on flow-matching generation that simultaneously preserves client privacy and resists sophisticated poisoning attacks. On the client side, each local network is decoupled into a private feature extractor and a public classifier. Each client is further equipped with a flow-matching generator that replaces the extractor when interacting with the server, thereby protecting private features while learning an approximation of the underlying data distribution. Complementing the client-side design, the server employs a client-update verification scheme and a novel robust aggregation mechanism driven by synthetic samples produced by the flow-matching generator. Experiments on MNIST, FMNIST, and CIFAR-10 demonstrate that, compared with prior work, our approach adapts to multiple attack strategies and achieves higher accuracy while maintaining strong privacy protection.
Read the full abstract

Cryptography and Security

Adversarial Attacks on Multimodal Large Language Models: A Comprehensive Survey

Bhavuk Jain, Sercan Ö. Ar{\i}k and Hardeo K. Thakur

Multimodal large language models (MLLMs) integrate information from multiple modalities such as text, images, audio, and video, enabling complex capabilities such as visual question answering and audio translation. While powerful, this increased expressiveness introduces new and amplified vulnerabilities to adversarial manipulation. This survey provides a comprehensive and systematic analysis of adversarial threats to MLLMs, moving beyond enumerating attack techniques to explain the underlying causes of model susceptibility. We introduce a taxonomy that organizes adversarial attacks according to attacker objectives, unifying diverse attack surfaces across modalities and deployment settings. Additionally, we also present a vulnerability-centric analysis that links integrity attacks, safety and jailbreak failures, control and instruction hijacking, and training-time poisoning to shared architectural and representational weaknesses in multimodal systems. Together, this framework provides an explanatory foundation for understanding adversarial behavior in MLLMs and informs the development of more robust and secure multimodal language systems.
Read the full abstract

Sunday 29 March

Cryptography and Security

Evaluating Prompt Injection Defenses for Educational LLM Tutors: Security-Usability-Latency Trade-offs

Alexandre Cristovão Maiorano

Educational LLM tutors face a core AI alignment challenge: they must follow user intent while preserving pedagogical constraints and safety policies. We present an evaluation methodology for prompt-injection defenses in this setting, showing that guardrail design entails explicit trade-offs among adversarial robustness, benign-task usability, and response latency. We evaluate a domain-specific multi-layer safeguard pipeline combining deterministic pattern filters, structural validation, contextual sandboxing, and session-level behavioral checks. On a controlled holdout benchmark, the pipeline reaches low bypass and false positive rates with optimized average latency - an operating point that prioritizes pedagogical usability (zero false positives) while maintaining measurable attack resistance. We provide a reproducible benchmark protocol for head-to-head comparison under identical conditions, including stratified bootstrap confidence intervals, paired McNemar significance tests, multi-seed sensitivity sweeps, and direct evaluation of Prompt Guard and NeMo Guardrails on the same split with unified instrumentation. Results expose operational trade-offs: NeMo reaches 0 percent bypass at 16.22 percent FPR and roughly 1.5s latency, while Prompt Guard yields 38.48 percent bypass with 3.60 percent FPR. The framework supports evidence-based guardrail selection for AI tutoring systems under different institutional risk and usability requirements.
Read the full abstract

Cryptography and Security

SNEAKDOOR: Stealthy Backdoor Attacks against Distribution Matching-based Dataset Condensation

He Yang, Dongyi Lv, Song Ma et al. (+2 more)

Dataset condensation aims to synthesize compact yet informative datasets that retain the training efficacy of full-scale data, offering substantial gains in efficiency. Recent studies reveal that the condensation process can be vulnerable to backdoor attacks, where malicious triggers are injected into the condensation dataset, manipulating model behavior during inference. While prior approaches have made progress in balancing attack success rate and clean test accuracy, they often fall short in preserving stealthiness, especially in concealing the visual artifacts of condensed data or the perturbations introduced during inference. To address this challenge, we introduce Sneakdoor, which enhances stealthiness without compromising attack effectiveness. Sneakdoor exploits the inherent vulnerability of class decision boundaries and incorporates a generative module that constructs input-aware triggers aligned with local feature geometry, thereby minimizing detectability. This joint design enables the attack to remain imperceptible to both human inspection and statistical detection. Extensive experiments across multiple datasets demonstrate that Sneakdoor achieves a compelling balance among attack success rate, clean test accuracy, and stealthiness, substantially improving the invisibility of both the synthetic data and triggered samples while maintaining high attack efficacy. The code is available at https://github.com/XJTU-AI-Lab/SneakDoor.
Read the full abstract

All authors He Yang, Dongyi Lv, Song Ma, Wei Xi, Jizhong Zhao

Cryptography and Security

Decentralized Proof-of-Location for Content Provenance: Towards Capture-Time Authenticity

Eduardo Brito, Fernando Castillo, Amnir Hadachi et al. (+2 more)

Reliable use of real-world data requires confidence that recorded evidence reflects what actually occurred at the moment of capture. In adversarial or incentive-misaligned cyber-physical settings, device-centric provenance and post-capture verification are insufficient to provide that guarantee. This paper builds on Proof-of-Location (PoL) as a baseline for establishing where and when events take place, and extends it with a witnessing-zone architecture in which multiple independent observers collectively validate physical events. The resulting approach produces auditable evidence artifacts that can support downstream systems in cyber-physical settings, without relying on centralized trust. Through representative scenarios and simulation-based evaluation, this paper shows how such architectures improve sensor data trustworthiness and resilience to fabricated or staged events.
Read the full abstract

All authors Eduardo Brito, Fernando Castillo, Amnir Hadachi, Ulrich Norbisrath, Jonathan Heiss

2603.27883 · 29 Mar · This work has been accepted for publication at the 5th International Workshop on Architecting and Engineering Digital Twins (AEDT 2026), to appear in the Companion Proceedings of the 23rd IEEE International Conference on Software Architecture (ICSA 2026)

Abstract PDF $ADVR ×1

Cryptography and Security · Cross-listed from cs.CV

Towards Context-Aware Image Anonymization with Multi-Agent Reasoning

Robert Aufschläger, Jakob Folz, Gautam Savaliya et al. (+3 more)

Street-level imagery contains personally identifiable information (PII), some of which is context-dependent. Existing anonymization methods either over-process images or miss subtle identifiers, while API-based solutions compromise data sovereignty. We present an agentic framework CAIAMAR (\underline{C}ontext-\underline{A}ware \underline{I}mage \underline{A}nonymization with \underline{M}ulti-\underline{A}gent \underline{R}easoning) for context-aware PII segmentation with diffusion-based anonymization, combining pre-defined processing for high-confidence cases with multi-agent reasoning for indirect identifiers. Three specialized agents coordinate via round-robin speaker selection in a Plan-Do-Check-Act (PDCA) cycle, enabling large vision-language models to classify PII based on spatial context (private vs. public property) rather than rigid category rules. The agents implement spatially-filtered coarse-to-fine detection where a scout-and-zoom strategy identifies candidates, open-vocabulary segmentation processes localized crops, and $IoU$-based deduplication ($30\%$ threshold) prevents redundant processing. Modal-specific diffusion guidance with appearance decorrelation substantially reduces re-identification (Re-ID) risks. On CUHK03-NP, our method reduces person Re-ID risk by $73\%$ ($R1$: $16.9\%$ vs. $62.4\%$ baseline). For image quality preservation on CityScapes, we achieve KID: $0.001$, and FID: $9.1$, significantly outperforming existing anonymization. The agentic workflow detects non-direct PII instances across object categories, and downstream semantic segmentation is preserved. Operating entirely on-premise with open-source models, the framework generates human-interpretable audit trails supporting EU's GDPR transparency requirements while flagging failed cases for human review.
Read the full abstract

All authors Robert Aufschläger, Jakob Folz, Gautam Savaliya, Manjitha D Vidanalage, Michael Heigl, Martin Schramm

Cryptography and Security

Ordering Power is Sanctioning Power: Sanction Evasion-MEV and the Limits of On-Chain Enforcement

Di Wu, Yuman Bai, Shoupeng Ren et al. (+5 more)

Centralized stablecoins such as USDT and USDC enforce sanctions through contract-layer blacklist functions. Yet on public blockchains, a freeze is still an ordinary transaction competing with the sanctioned party's transfer for priority. It exposes a gap between contract-layer authority and ordering-layer enforcement: when both race for the same block, the outcome is set not by legal mandate, but by block producers' choices. Because both sides can pay for priority, sanction races create rents for block producers, which we call Sanction-Evasion MEV (SE-MEV). To measure this gap, we build the first longitudinal dataset of on-chain sanctions enforcement and evasion for Ethereum-based USDT and USDC from November 2017 to August 2025, covering more than $1.5 billion in frozen value. At least 7.3% of sanctioned USDT addresses and 18.7% of sanctioned USDC addresses had already been drained to zero before the freeze took effect. We also trace an escalation from issuer-side out-of-gas failures, to public gas auctions, private order flow, and direct payments to block producers, showing that block producers extract MEV from sanction enforcement. We then develop a game-theoretic model of stablecoin sanctions with MEV. It shows that compliant issuers cannot rationally stay outside the ordering market; fixed participation costs concentrate evasion among specialized MEV-aware adversaries; and the implicit MEV tax rises with regulatory penalties, creating incentives for vertical integration into block-building infrastructure. The problem extends beyond stablecoins. Any privileged on-chain action executed as an ordinary transaction – emergency pauses, governance interventions, or judicial freezes – faces the same conflict. Where ordering power follows economic incentives, ordering power is sanctioning power; contract-layer authority alone cannot guarantee enforcement.
Read the full abstract

All authors Di Wu, Yuman Bai, Shoupeng Ren, Xinyu Zhang, Yiyue Cao, Xuechao Wang, Wu Wen, Jian Liu

Cryptography and Security · Cross-listed from cs.CL

Hidden Ads: Behavior Triggered Semantic Backdoors for Advertisement Injection in Vision Language Models

Duanyi Yao, Changyue Li, Zhicong Huang et al. (+2 more)

Vision-Language Models (VLMs) are increasingly deployed in consumer applications where users seek recommendations about products, dining, and services. We introduce Hidden Ads, a new class of backdoor attacks that exploit this recommendation-seeking behavior to inject unauthorized advertisements. Unlike traditional pattern-triggered backdoors that rely on artificial triggers such as pixel patches or special tokens, Hidden Ads activates on natural user behaviors: when users upload images containing semantic content of interest (e.g., food, cars, animals) and ask recommendation-seeking questions, the backdoored model provides correct, helpful answers while seamlessly appending attacker-specified promotional slogans. This design preserves model utility and produces natural-sounding injections, making the attack practical for real-world deployment in consumer-facing recommendation services. We propose a multi-tier threat framework to systematically evaluate Hidden Ads across three adversary capability levels: hard prompt injection, soft prompt optimization, and supervised fine-tuning. Our poisoned data generation pipeline uses teacher VLM-generated chain-of-thought reasoning to create natural trigger–slogan associations across multiple semantic domains. Experiments on three VLM architectures demonstrate that Hidden Ads achieves high injection efficacy with near-zero false positives while maintaining task accuracy. Ablation studies confirm that the attack is data-efficient, transfers effectively to unseen datasets, and scales to multiple concurrent domain-slogan pairs. We evaluate defenses including instruction-based filtering and clean fine-tuning, finding that both fail to remove the backdoor without causing significant utility degradation.
Read the full abstract

All authors Duanyi Yao, Changyue Li, Zhicong Huang, Cheng Hong, Songze Li

Cryptography and Security

A Security Analysis of the OpenClaw AI Agent Framework

Surada Suwansathit, Yuxuan Zhang and Guofei Gu

AI agent frameworks connecting large language model (LLM) reasoning to host execution surfaces – shell, filesystem, containers, and messaging – introduce security challenges structurally distinct from conventional software. We present a systematic taxonomy of 470 advisories filed against OpenClaw, an open-source AI agent runtime, organized by architectural layer and trust-violation type. Vulnerabilities cluster along two orthogonal axes: (1) the system axis, reflecting the architectural layer (exec policy, gateway, channel, sandbox, browser, plugin, agent/prompt); and (2) the attack axis, reflecting adversarial techniques (identity spoofing, policy bypass, cross-layer composition, prompt injection, supply-chain escalation). Patch-differential evidence yields three principal findings. First, three Moderate- or High-severity advisories in the Gateway and Node-Host subsystems compose into a complete unauthenticated remote code execution (RCE) path – spanning delivery, exploitation, and command-and-control – from an LLM tool call to the host process. Second, the exec allowlist, the primary command-filtering mechanism, relies on a closed-world assumption that command identity is recoverable via lexical parsing. This is invalidated by shell line continuation, busybox multiplexing, and GNU option abbreviation. Third, a malicious skill distributed via the plugin channel executed a two-stage dropper within the LLM context, bypassing the exec pipeline and demonstrating that the skill distribution surface lacks runtime policy enforcement. The dominant structural weakness is per-layer trust enforcement rather than unified policy boundaries, making cross-layer attacks resilient to local remediation.
Read the full abstract

Saturday 28 March

Cryptography and Security

Safety in Embodied AI: A Survey of Risks, Attacks, and Defenses

Xiao Li, Xiang Zheng, Yifeng Gao et al. (+35 more)

Embodied Artificial Intelligence (Embodied AI) integrates perception, cognition, planning, and interaction into agents that operate in open-world, safety-critical environments. As these systems gain autonomy and enter domains such as transportation, healthcare, and industrial or assistive robotics, ensuring their safety becomes both technically challenging and socially indispensable. Unlike digital AI systems, embodied agents must act under uncertain sensing, incomplete knowledge, and dynamic human-robot interactions, where failures can directly lead to physical harm. This survey provides a comprehensive and structured review of safety research in embodied AI, examining attacks and defenses across the full embodied pipeline, from perception and cognition to planning, action and interaction, and agentic system. We introduce a multi-level taxonomy that unifies fragmented lines of work and connects embodied-specific safety findings with broader advances in vision, language, and multimodal foundation models. Our review synthesizes insights from over 500 papers spanning adversarial, backdoor, jailbreak, and hardware-level attacks; attack detection, safe training and robust inference; and risk-aware human-agent interaction. This analysis reveals several overlooked challenges, including the fragility of multimodal perception fusion, the instability of planning under jailbreak attacks, and the trustworthiness of human-agent interaction in open-ended scenarios. By organizing the field into a coherent framework and identifying critical research gaps, this survey provides a roadmap for building embodied agents that are not only capable and autonomous but also safe, robust, and reliable in real-world deployment.
Read the full abstract

All authors Xiao Li, Xiang Zheng, Yifeng Gao, Xinyu Xia, Yixu Wang, Xin Wang, Ye Sun, Yunhan Zhao, Ming Wen, Jiayu Li, Zixing Chen, Xun Gong, Yi Liu, Yige Li, Yutao Wu, Cong Wang, Jun Sun, Yixin Cao, Zhineng Chen, Jingjing Chen, Tao Gui, Qi Zhang, Zuxuan Wu, Xipeng Qiu, Xuanjing Huang, Tiehua Zhang, Zhipeng Wei, Kun Wang, Xinfeng Li, Hanxun Huang, Sarah Erfani, James Bailey, Jianping Wang, Chaowei Xiao, Ran He, Bo Li, Xingjun Ma, Yu-Gang Jiang

Cryptography and Security

GUARD-SLM: Token Activation-Based Defense Against Jailbreak Attacks for Small Language Models

Md Jueal Mia, Joaquin Molto, Yanzhao Wu et al. (+1 more)

Small Language Models (SLMs) are emerging as efficient and economically viable alternatives to Large Language Models (LLMs), offering competitive performance with significantly lower computational costs and latency. These advantages make SLMs suitable for resource-constrained and efficient deployment on edge devices. However, existing jailbreak defenses show limited robustness against heterogeneous attacks, largely due to an incomplete understanding of the internal representations across different layers of language models that facilitate jailbreak behaviors. In this paper, we conduct a comprehensive empirical study on 9 jailbreak attacks across 7 SLMs and 3 LLMs. Our analysis shows that SLMs remain highly vulnerable to malicious prompts that bypass safety alignment. We analyze hidden-layer activations across different layers and model architectures, revealing that different input types form distinguishable patterns in the internal representation space. Based on this observation, we propose GUARD-SLM, a lightweight token activation-based method that operates in the representation space to filter malicious prompts during inference while preserving benign ones. Our findings highlight robustness limitations across layers of language models and provide a practical direction for secure small language model deployment.
Read the full abstract

All authors Md Jueal Mia, Joaquin Molto, Yanzhao Wu, M. Hadi Amini