Papers

As large language models (LLMs) are increasingly trained on sensitive user data, understanding the fundamental cost of privacy in language learning becomes essential. We initiate the study of differentially private (DP) language identification and generation in the agnostic statistical setting, establishing algorithms and matching lower bounds that precisely quantify the cost of privacy. For both tasks, approximate $(\varepsilon, \delta)$-DP with constant $\varepsilon > 0$ recovers the non-private error rates: $\exp(-r(n))$ for identification (for any $r(n) = o(n)$) and $\exp(-\Omega(n))$ for generation. Under pure $\varepsilon$-DP, the exponents degrade by a multiplicative factor of $\min\{1, \varepsilon\}$, which we show is tight up to constants. Notably, for generation under pure DP with mild assumptions, the upper bound $\exp(-\min\{1,\varepsilon\} \cdot \Omega(n))$ matches the lower bound up to some constants, establishing an optimal rate. Our results show that the cost of privacy in language learning is surprisingly mild: absent entirely under approximate DP, and exactly a $\min\{1,\varepsilon\}$ factor in the exponent under pure DP.

As large language models (LLMs) evolve from static chatbots into autonomous agents, the primary vulnerability surface shifts from final outputs to intermediate execution traces. While safety guardrails are well-benchmarked for natural language responses, their efficacy remains largely unexplored within multi-step tool-use trajectories. To address this gap, we introduce **TraceSafe-Bench**, the first comprehensive benchmark specifically designed to assess mid-trajectory safety. It encompasses 12 risk categories, ranging from security threats (e.g., prompt injection, privacy leaks) to operational failures (e.g., hallucinations, interface inconsistencies), featuring over 1,000 unique execution instances. Our evaluation of 13 LLM-as-a-guard models and 7 specialized guardrails yields three critical findings: 1) *Structural Bottleneck*: Guardrail efficacy is driven more by structural data competence (e.g., JSON parsing) than semantic safety alignment. Performance correlates strongly with structured-to-text benchmarks ($\rho=0.79$, $p<0.01$) while we detect no association with standard jailbreak robustness. 2) *Scale Is Not the Deciding Factor*: Trajectory detection accuracy does not scale monotonically with model size, and general-purpose LLMs consistently outperform specialized safety guardrails. 3) *Bounded Temporal Stability*: Detection accuracy improves with trajectory length within native context bounds as models process dynamic execution behavior, but degrades in extreme long-context regimes due to over-refusal.

This article presents DDP-SA, a scalable privacy-preserving federated learning framework that jointly leverages client-side local differential privacy (LDP) and full-threshold additive secret sharing (ASS) for secure aggregation. Unlike existing methods that rely solely on differential privacy or on secure multi-party computation (MPC), DDP-SA integrates both techniques to deliver stronger end-to-end privacy guarantees while remaining computationally practical. The framework introduces a two-stage protection mechanism: clients first perturb their local gradients with calibrated Laplace noise, then decompose the noisy gradients into additive secret shares that are distributed across multiple intermediate servers. This design ensures that (i) no single compromised server or communication channel can reveal any information about individual client updates, and (ii) the parameter server reconstructs only the aggregated noisy gradient, never any client-specific contribution. Extensive experiments show that DDP-SA achieves substantially higher model accuracy than standalone LDP while providing stronger privacy protection than MPC-only approaches. The proposed framework scales linearly with the number of participants and offers a practical, privacy-preserving solution for federated learning applications with controllable computational and communication overhead.

Touch-based authentication is widely deployed on mobile devices due to its convenience and seamless user experience. However, existing systems largely model touch interaction as a purely behavioral signal, overlooking its intrinsic multidimensional nature and limiting robustness against sophisticated adversarial behaviors and real-world variations. In this work, we present BioMoTouch, a multi-modal touch authentication framework on mobile devices grounded in a key empirical finding: during touch interaction, inertial sensors capture user-specific behavioral dynamics, while capacitive screens simultaneously capture physiological characteristics related to finger morphology and skeletal structure. Building upon this insight, BioMoTouch jointly models physiological contact structures and behavioral motion dynamics by integrating capacitive touchscreen signals with inertial measurements. Rather than combining independent decisions, the framework explicitly learns their coordinated interaction to form a unified representation of touch behavior. BioMoTouch operates implicitly during natural user interactions and requires no additional hardware, enabling practical deployment on commodity mobile devices. We evaluate BioMoTouch with 38 participants under realistic usage conditions. Experimental results show that BioMoTouch achieves a balanced accuracy of 99.71% and an equal error rate of 0.27%. Moreover, it maintains false acceptance rates below 0.90% under artificial replication, mimicry, and puppet attack scenarios, demonstrating strong robustness against partial-factor manipulation.

Palmprint recognition is deployed in security-critical applications, including access control and palm-based payment, due to its contactless acquisition and highly discriminative ridge-and-crease textures. However, the robustness of deep palmprint recognition systems against physically realizable attacks remains insufficiently understood. Existing studies are largely confined to the digital setting and do not adequately account for the texture-dominant nature of palmprint recognition or the distortions introduced during physical acquisition. To address this gap, we propose CAAP, a capture-aware adversarial patch framework for palmprint recognition. CAAP learns a universal patch that can be reused across inputs while remaining effective under realistic acquisition variation. To match the structural characteristics of palmprints, the framework adopts a cross-shaped patch topology, which enlarges spatial coverage under a fixed pixel budget and more effectively disrupts long-range texture continuity. CAAP further integrates three modules: ASIT for input-conditioned patch rendering, RaS for stochastic capture-aware simulation, and MS-DIFE for feature-level identity-disruptive guidance. We evaluate CAAP on the Tongji, IITD, and AISEC datasets against generic CNN backbones and palmprint-specific recognition models. Experiments show that CAAP achieves strong untargeted and targeted attack performance with favorable cross-model and cross-dataset transferability. The results further show that, although adversarial training can partially reduce the attack success rate, substantial residual vulnerability remains. These findings indicate that deep palmprint recognition systems remain vulnerable to physically realizable, capture-aware adversarial patch attacks, underscoring the need for more effective defenses in practice. Code available at https://github.com/ryliu68/CAAP.

With the rapid advancement of decentralized applications, smart contract security faces severe challenges, particularly regarding atomicity violations in complex logic such as Oracle and NFT contracts. Rigid rule sets often limit traditional static analyzers and lack deep contextual awareness, leading to high false-positive and false-negative rates when identifying vulnerabilities that depend on intermediate state inconsistencies. To address these limitations, this paper proposes PSR\textsuperscript{2}, a novel collaborative static analysis framework that integrates structural path searching with deterministic semantic reasoning. PSR\textsuperscript{2} utilizes a Graph Structure Analysis Module (GSAM) to identify suspicious execution sequences in control flow graphs and a Semantic Context Analysis Module (SCAM) to extract data dependencies and state facts from abstract syntax trees. A Fusion Decision Module (FDM) then performs formal cross validation to confirm vulnerabilities based on a unified atomicity inconsistency model. Experimental results on 1,600 contract samples demonstrate that PSR\textsuperscript{2} significantly outperforms pattern-matching baselines, achieving an F1-score of 92.36% in complex ERC-721 scenarios compared to 51.86% for existing tools. Ablation studies further confirm that our fusion logic effectively reduces the false-positive rate by nearly half compared to single module analysis.

The rapid growth of software vulnerabilities has turned cyber threat intelligence analysis into a challenging data mining problem over heterogeneous and continuously changing sources. Public repositories such as the National Vulnerability Database (NVD), Common Vulnerabilities and Exposures (CVE), Common Weakness Enumeration (CWE), Exploit Database (EDB), and CVE Details provide valuable information, but their record-centric schemas make it difficult to capture cross-source relationships among vulnerabilities, weaknesses, exploits, affected products, vendors, and references. Existing graph-based vulnerability resources highlight the value of relational threat modelling, yet many remain static, offline, or difficult to access for downstream graph mining. This paper presents VulLink, a deployed, dynamic, and open-access vulnerability graph database for cybersecurity data mining. VulLink integrates multiple public repositories through an automated Extract-Transform-Load (ETL) pipeline that converts isolated, record-centric vulnerability data into a continuously updated graph database with typed entities and explicit cross-source relationships. It provides an interactive Web interface and public API for exploring, querying, and exporting mining-ready vulnerability subgraphs. It also provides pre-computed embeddings of vulnerability descriptions generated by pretrained language models, which users can query and download by model and embedding dimension as semantic features for downstream mining tasks such as exploitability prediction. To demonstrate the practical utility of VulLink, we implement a downstream exploitability prediction use case that leverages heterogeneous graph context and semantic vulnerability features. The VulLink platform, including the Web interface, public API, source code, and deployment resources, is publicly available online.

Ensuring ciphertext indistinguishability is fundamental to cryptographic security, but empirically validating this property in real implementations and hybrid settings presents practical challenges. The transition to post-quantum cryptography (PQC), with its hybrid constructions combining classical and quantum-resistant primitives, makes empirical validation approaches increasingly valuable. By modeling IND-CPA games as binary classification tasks and training on labeled ciphertext data with BCE loss, we study deep neural network (DNN) distinguishers for ciphertext indistinguishability. We apply this methodology to PQC KEMs. We specifically test the public-key encryption (PKE) schemes used to construct examples such as ML-KEM, BIKE, and HQC. Moreover, a novel extension of this DNN modeling for empirical distinguishability testing of hybrid KEMs is presented. We implement and test this on combinations of PQC KEMs with plain RSA, RSA-OAEP, and plaintext. Finally, methodological generality is illustrated by applying the DNN IND-CPA classification framework to cascade symmetric encryption, where we test combinations of AES-CTR, AES-CBC, AES-ECB, ChaCha20, and DES-ECB. In our experiments on PQC algorithms, KEM combiners, and cascade encryption, no algorithm or combination of algorithms demonstrates a significant advantage (two-sided binomial test, significance level $\alpha = 0.01$), consistent with theoretical guarantees that hybrids including at least one IND-CPA-secure component preserve indistinguishability, and with the absence of exploitable patterns under the considered DNN adversary model. These illustrate the potential of using deep learning as an adaptive, practical, and versatile empirical estimator for indistinguishability in more general IND-CPA settings, allowing data-driven validation of implementations and compositions and complementing the analytical security analysis.

The field of cybersecurity is confronted with two interrelated challenges: a worldwide deficit of qualified practitioners and ongoing human-factor weaknesses that account for the bulk of security incidents. To tackle these issues, we present SentinelSphere, a platform driven by artificial intelligence that unifies machine learning-based threat identification with security training powered by a Large Language Model (LLM). The detection module uses an Enhanced Deep Neural Network (DNN) trained on the CIC-IDS2017 and CIC-DDoS2019 benchmark datasets, enriched with novel HTTP-layer feature engineering that captures application level attack signatures. For the educational component, we deploy a quantised variant of Phi-4 model (Q4_K_M), fine-tuned for the cybersecurity domain, enabling deployment on commodity hardware requiring only 16 GB of RAM without dedicated GPU resources. Experimental results show that the Enhanced DNN attains high detection accuracy while substantially lowering false positives relative to baseline models, and maintains strong recall across critical attack categories such as DDoS, brute force, and web-based exploits. Validation workshops involving industry professionals and university students confirmed that the Traffic Light visualisation system and conversational AI assistant are both intuitive and effective for users without technical backgrounds. SentinelSphere illustrates that coupling intelligent threat detection with adaptive, LLM-driven security education can meaningfully address both technical and human-factor cybersecurity vulnerabilities within a single, cohesive framework.

Data leakage is the inadvertent transfer of information between training and evaluation datasets that poses a subtle, yet critical, risk to the reliability of machine learning (ML) models in safety-critical systems such as automotive perception. While leakage is widely recognized in research, little is known about how industrial practitioners actually perceive and manage it in practice. This study investigates practitioners' knowledge, experiences, and mitigation strategies around data leakage through ten semi-structured interviews with system design, development, and verification engineers working on automotive perception functions development. Using reflexive thematic analysis, we identify that knowledge of data leakage is widespread and fragmented along role boundaries: ML engineers conceptualize it as a data-splitting or validation issue, whereas design and verification roles interpret it in terms of representativeness and scenario coverage. Detection commonly arises through generic considerations and observed performance anomalies rather than implying specific tools. However, data leakage prevention is more commonly practiced, which depends mostly on experience and knowledge sharing. These findings suggest that leakage control is a socio-technical coordination problem distributed across roles and workflows. We discuss implications for ML reliability engineering, highlighting the need for shared definitions, traceable data practices, and continuous cross-role communication to institutionalize data leakage awareness within automotive ML development.

While Chain-of-Thought (CoT) prompting has become a standard paradigm for eliciting complex reasoning capabilities in Large Language Models, it inadvertently exposes a new attack surface for backdoor attacks. Existing CoT backdoor attacks typically manipulate the intermediate reasoning steps to steer the model toward incorrect answers. However, these corrupted reasoning traces are readily detected by prevalent process-monitoring defenses. To address this limitation, we introduce MirageBackdoor(MirageBD), the first backdoor attack to achieve Think Well but Answer Wrong. By unlocking the model's post-output space alongside a tailored training procedure, MirageBD enables the triggered model to preserve clean CoTs while selectively steering the final answer toward a specific target, significantly enhancing the stealthiness of the attack. Experiments show that MirageBD generally achieves over 90% attack success rate across four datasets and five models with a poison ratio of only 5%. Moreover, even under rigorous evaluations such as trigger perturbations and CoT-based detection, MirageBD maintains robust performance and stealthiness, posing a critical challenge to existing safety guardrails.

As high quality public data becomes scarce, Federated Learning (FL) provides a vital pathway to leverage valuable private user data while preserving privacy. However, real-world client data often contains toxic or unsafe information. This leads to a critical issue we define as unintended data poisoning, which can severely damage the safety alignment of global models during federated alignment. To address this, we propose FedDetox, a robust framework tailored for Small Language Models (SLMs) on resource-constrained edge devices. We first employ knowledge distillation to transfer sophisticated safety alignment capabilities from large scale safety aligned teacher models into light weight student classifiers suitable for resource constrained edge devices. Specifically, during federated learning for human preference alignment, the edge client identifies unsafe samples at the source and replaces them with refusal templates, effectively transforming potential poisons into positive safety signals. Experiments demonstrate that our approach preserves model safety at a level comparable to centralized baselines without compromising general utility.

Current LLM-based services typically require users to submit raw text regardless of its sensitivity. While intuitive, such practice introduces substantial privacy risks, as unauthorized access may expose personal, medical, or legal information. Although prior defenses strived to mitigate these risks, they often incur substantial computational overhead and degrade model performance. To overcome this privacy-efficiency trade-off, we introduce Privacy-Preserving Fine-Tuning (PPFT), a novel training pipeline that eliminates the need for transmitting raw prompt text while maintaining a favorable balance between privacy preservation and model utility for both clients and service providers. Our approach operates in two stages: first, we train a client-side encoder together with a server-side projection module and LLM, enabling the server to condition on k-pooled prompt embeddings instead of raw text; second, we fine-tune the projection module and LLM on private, domain-specific data using noise-injected embeddings, allowing effective adaptation without exposing plain text prompts and requiring access to the decoder's internal parameters. Extensive experiments on domain-specific and general benchmarks demonstrate that PPFT achieves a striking balance between privacy and utility, maintaining competitive performance with minimal degradation compared to noise-free upper bounds.

Skill-based agent systems tackle complex tasks by composing reusable skills, improving modularity and scalability while introducing a largely unexamined security attack surface. We propose SkillTrojan, a backdoor attack that targets skill implementations rather than model parameters or training data. SkillTrojan embeds malicious logic inside otherwise plausible skills and leverages standard skill composition to reconstruct and execute an attacker-specified payload. The attack partitions an encrypted payload across multiple benign-looking skill invocations and activates only under a predefined trigger. SkillTrojan also supports automated synthesis of backdoored skills from arbitrary skill templates, enabling scalable propagation across skill-based agent ecosystems. To enable systematic evaluation, we release a dataset of 3,000+ curated backdoored skills spanning diverse skill patterns and trigger-payload configurations. We instantiate SkillTrojan in a representative code-based agent setting and evaluate both clean-task utility and attack success rate. Our results show that skill-level backdoors can be highly effective with minimal degradation of benign behavior, exposing a critical blind spot in current skill-based agent architectures and motivating defenses that explicitly reason about skill composition and execution. Concretely, on EHR SQL, SkillTrojan attains up to 97.2% ASR while maintaining 89.3% clean ACC on GPT-5.2-1211-Global.

Security Information and Event Management (SIEM) systems make it possible for detecting intrusion anomalies in real-time manner by their applied security rules. However, the heterogeneity of vendor-specific rules (e.g., Splunk SPL, Microsoft KQL, IBM AQL, Google YARA-L, and RSA ESA) makes cross-platform rule reuse extremely difficult, requiring deep domain knowledge for reliable conversion. As a result, an autonomous and accurate rule conversion framework can significantly lead to effort savings, preserving the value of existing rules. In this paper, we propose ARuleCon, an agentic SIEM-rule conversion approach. Using ARuleCon, the security professionals do not need to distill the source rules' logic, the documentation of the target rules and ARuleCon can purposely convert to the target vendors without more intervention. To achieve this, ARuleCon is equipped with conversion/schema mismatches, and Python-based consistency check that running both source and target rules in controlled test environments to mitigate subtle semantic drifts. We present a comprehensive evaluation of ARuleCon ranging from textual alignment and the execution success, showcasing ARuleCon can convert rules with high fidelity, outperforming the baseline LLM model by 15% averagely. Finally, we perform case studies and interview with our industry collaborators in Singtel Singapore, which showcases that ARuleCon can significantly save expert's time on understanding cross-SIEM's documentation and remapping logic.

The lead marketing ecosystem enables collection, sale, and use of personal data submitted via web forms to deliver personalized quotes in high-value verticals such as insurance. Despite its scale and sensitivity of the collected data, this ecosystem remains largely unexplored by the research community. We present the first empirical study of privacy and spam risks in lead marketing, developing an end-to-end measurement framework to trace data flows from data collection to consumer contact. Our setup instruments over 100 health-related lead-generation websites and monitors 200 controlled phone numbers and email addresses to understand downstream marketing practices. We observe sharing of highly personal and sensitive health information to more than 70 distinct third parties on these lead generation websites. By purchasing our own and other organic leads from three major lead platforms, we uncover deceptive brokerage practices, where consumer data is sold to unvetted buyers and often augmented or fabricated with attributes such as health status and weight. We received a total of over 8,000 telemarketing phone calls, 600 text messages, and 200 emails, where calls often began within seconds of form submission. Many campaigns relied on VoIP-based neighbor spoofing and high-frequency dialing, at times rendering phones unusable. Our experiments with phone and email opt-outs suggest phone-based opt-outs to help the most, although all were ineffective at completely stopping marketing communications. Analysis of 7,432 Better Business Bureau (BBB) complaints and reviews corroborates these findings from the consumer perspective. Overall, our results reveal a highly interconnected and non-compliant lead marketing ecosystem that aggressively monetizes sensitive consumer data.

In video conferencing, human faces serve as the primary visual focal points, playing multifaceted roles that enhance visual communication and emotional connection. However, we argue that a human face is also a side channel, which can unwittingly leak on-screen information through online video feeds. To demonstrate this, we conduct feasibility studies, which reveal that, illuminated by both ambient light and light emitted from displays, the human face can reflect optical variations of different on-screen content. The paper then proposes FaceTell, a novel side-channel attack system that eavesdrops on fine-grained application activities from pervasive yet subtle facial reflections during video conferencing. We implement FaceTell in a real-world testbed with three different brands of laptops and four mainstream video conferencing platforms. FaceTell is then evaluated with 24 human subjects across 13 unique indoor environments. With more than 12 hours of video data, FaceTell achieves a high accuracy of 99.32% for eavesdropping on 28 popular applications and is resilient to many practical impact factors. Finally, potential countermeasures are proposed to mitigate this new attack.

Quantum computing simulators form the classical software foundation on which virtually all quantum algorithm research depends. We present Broken Quantum, the first comprehensive formal security audit of the open-source quantum computing simulator ecosystem. Applying COBALT QAI – a four-module static analysis engine backed by the Z3 SMT solver – we analyze 45 open-source quantum simulation frameworks from 22 organizations spanning 12 countries. We identify 547 security findings (40 CRITICAL, 492 HIGH, 15 MEDIUM) across four vulnerability classes: CWE-125/190 (C++ memory corruption), CWE-400 (Python resource exhaustion), CWE-502/94 (unsafe deserialization and code injection), and CWE-77/22 (QASM injection – a novel, quantum-specific attack vector with no classical analog). All 13 vulnerability patterns are formally verified via Z3 satisfiability proofs (13/13 SAT). The 32-qubit boundary emerges as a consistent formal threshold in both C++ and Python vulnerability chains. Supply chain analysis identifies the first documented case of vulnerability transfer from a commercial quantum framework into US national laboratory infrastructure (IBM Qiskit Aer to XACC/Oak Ridge National Laboratory). Nine frameworks score 100/100 under all four scanners; Qiskit Aer,Cirq, tequila, PennyLane, and 5 others score 0/100.

Recent standards such as RSL address AI content policy declaration – telling AI systems what the licensing terms are. However, no existing system provides audit infrastructure – tamper-evident licensing transaction records with independently verifiable proofs that those records have not been retroactively modified. We describe Aegon, a protocol that extends standard JWT tokens with content-specific licensing claims and maintains a Certificate Transparency-style Merkle tree over an append-only transaction ledger, enabling third-party auditors to independently verify that specific content licensing transactions were recorded and have not been retroactively modified. Publishers validate tokens at the edge using standard JWKS with no broker dependency in the content delivery path. A signed provenance event log tracks content through AI transformation stages (chunking, embedding, retrieval, citation), bound to ledger entries by transaction ID. We further describe hardware-attested compliance receipts for on-device Android AI agents using StrongBox secure element attestation – to our knowledge, the first application of hardware-attested compliance receipts to AI content licensing. Existing DRM systems use hardware-backed keys for content decryption but do not produce verifiable compliance receipts for audit trails. We describe a reference architecture and an evaluation methodology for measuring protocol overhead. The protocol runs entirely over standard HTTPS and is designed to complement existing licensing standards rather than replace them.

Effective detection of unknown network security threats in multi-class imbalanced environments is critical for maintaining cyberspace security. Current methods focus on learning class representations but face challenges with unknown threat detection, class imbalance, and lack of interpretability, limiting their practical use. To address this, we propose RPM-Net, a novel framework that introduces reciprocal point mechanism to learn "non-class" representations for each known attack category, coupled with adversarial margin constraints that provide geometric interpretability for unknown threat detection. RPM-Net++ further enhances performance through Fisher discriminant regularization. Experimental results show that RPM-Net achieves superior performance across multiple metrics including F1-score, AUROC, and AUPR-OUT, significantly outperforming existing methods and offering practical value for real-world network security applications. Our code is available at:https://github.com/chiachen-chang/RPM-Net

Recent advancements in Large Language Models (LLMs) have sparked interest in their application to Static Application Security Testing (SAST), primarily due to their superior contextual reasoning capabilities compared to traditional symbolic or rule-based methods. However, existing LLM-based approaches typically attempt to replace human experts directly without integrating effectively with existing SAST tools. This lack of integration results in ineffectiveness, including high rates of false positives, hallucinations, limited reasoning depth, and excessive token usage, making them impractical for industrial deployment. To overcome these limitations, we present a paradigm shift that reorchestrates the SAST workflow from current LLM-assisted structure to a new LLM-centered workflow. We introduce Argus (Agentic and Retrieval-Augmented Guarding System), the first multi-agent framework designed specifically for vulnerability detection. Argus incorporates three key novelties: comprehensive supply chain analysis, collaborative multi-agent workflows, and the integration of state-of-the-art techniques such as Retrieval-Augmented Generation (RAG) and ReAct to minimize hallucinations and enhance reasoning. Extensive empirical evaluation demonstrates that Argus significantly outperforms existing methods by detecting a higher volume of true vulnerabilities while simultaneously reducing false positives and operational costs. Notably, Argus has identified several critical zero-day vulnerabilities with CVE assignments.

While recent approaches leverage large language models (LLMs) and multi-agent pipelines to automatically generate proof-of-concept (PoC) exploits from vulnerability reports, existing systems often suffer from two fundamental limitations: unreliable validation based on surface-level execution signals and high operational cost caused by extensive trial-and-error during exploit generation. In this paper, we present PoC-Adapt, an end-to-end framework for automated PoC generation and verification, architected upon a foundation semantic runtime validation and adaptive policy learning. At the core of PoC-Adapt is a Semantic Oracle that validates exploits by comparing structured pre- and post-execution system states, enabling reliable distinction between true vulnerability exploitation and incidental behavioral changes. To reduce exploration cost, we further introduce an Adaptive Policy Learning mechanism that learns an exploitation policy over semantic states and actions, guiding the exploit agent toward effective strategies with fewer failed attempts. PoC-Adapt is implemented as a multi-agent system comprising specialized agents for root cause analysis, environment building, exploit generation, and semantic validation, coordinated through structured feedback loops. Experimenting on the CWE-Bench-Java and PrimeVul benchmarks shows that PoC-Adapt significantly improves verification reliability by 25% and reduces exploit generation cost compared to prior LLM-based systems, highlighting the importance of semantic validation and learned action policies in automated vulnerability reproduction. Applied to the latest CVE corpus, PoC-Adapt confirmed 12 verified PoC out of 80 reproduce attempts at a cost of $0.42 per generated exploit

Concept drift and adversarial evasion are two major challenges for deploying machine learning-based malware detectors. While both have been studied separately, their combination, the adversarial robustness of drift-adaptive detectors, remains unexplored. We address this problem with AdvDA, a recent malware detector that uses adversarial domain adaptation to align a labeled source domain with a target domain with limited labels. The distribution shift between domains poses a unique challenge: robustness learned on the source may not transfer to the target, and existing defenses assume a fixed distribution. To address this, we propose a universal robustification framework that fine-tunes a pretrained AdvDA model on adversarially transformed inputs, agnostic to the attack type and choice of transformations. We instantiate it with five defense variants spanning two threat models: white-box PGD attacks in the feature space and black-box MalGuise attacks that modify malware binaries via functionality-preserving control-flow mutations. Across nine defense configurations, five monthly adaptation windows on Windows malware, and three false-positive-rate operating points, we find the undefended AdvDA completely vulnerable to PGD (100% attack success) and moderately to MalGuise (13%). Our framework reduces these rates to as low as 3.2% and 5.1%, respectively, but the optimal strategy differs: source adversarial training is essential for PGD defenses yet counterproductive for MalGuise defenses, where target-only training suffices. Furthermore, robustness does not transfer across these two threat models. We provide deployment recommendations that balance robustness, detection accuracy, and computational cost.

Agent skills combine natural-language instructions with executable code while inheriting an agent's filesystem, credential, and network access. Attacks can span prose and files, whereas regex and code-only analyzers cover only one modality. SkillSieve applies three progressively deeper layers: recall-oriented regex, AST, and metadata triage; four parallel LLM security sub-tasks; and an independent three-model jury with debate on disagreement. We evaluate 49,592 real ClawHub skills, a 390-skill labeled benchmark, and 100 adversarial samples across five evasion techniques on a 440 USD ARM board. The full pipeline achieves F1 = 0.929 (precision 0.912, recall 0.945) at an average cost of $0.006 per skill. An optional XGBoost fast path reduces Layer-2/3 calls by 32% with a 1.7-point F1 decrease while preserving recall. On 52 Feishu/Lark packages, Layer 2 reclassifies 13 of 14 Layer-1 flags as safe after contextual analysis; we also deploy the system as a Feishu chat bot. Code, labels, and aggregate results are open-sourced.

Symbolic execution detects vulnerabilities with precision, but applying it to large codebases requires harnesses that set up symbolic state, model dependencies, and specify assertions. Writing these harnesses has traditionally been a manual process requiring expert knowledge, which significantly limits the scalability of the technique. We present Static Analysis Informed and LLM-Orchestrated Symbolic Execution (SAILOR), which automates symbolic execution harness construction by combining static analysis with LLM-based synthesis. SAILOR operates in three phases: (1) static analysis identifies candidate vulnerable locations and generates vulnerability specifications; (2) an LLM uses vulnerability specifications and orchestrates harness synthesis by iteratively refining drivers, stubs, and assertions against compiler and symbolic execution feedback; symbolic execution then detects vulnerabilities using the generated harness, and (3) concrete replay validates the symbolic execution results against the unmodified project source. This design combines the scalability of static analysis, the code reasoning of LLMs, the path precision of symbolic execution, and the ground truth produced by concrete execution. We evaluate SAILOR on 10 open-source C/C++ projects totaling 6.8 M lines of code. SAILOR discovers 379 distinct, previously unknown memory-safety vulnerabilities (421 confirmed crashes). The strongest of five baselines we compare SAILOR to (agentic vulnerability detection using Claude Code with full codebase access and unlimited interaction), finds only 12 vulnerabilities. Each phase of SAILOR is critical: Without static analysis targeting confirmed vulnerabilities drop 12.2X; without iterative LLM synthesis zero vulnerabilities are confirmed; and without symbolic execution no approach can detect more than 12 vulnerabilities.

We study stochastic convex optimization (SCO) with heavy-tailed gradients under pure $\varepsilon$-differential privacy (DP). Instead of assuming a bound on the worst-case Lipschitz parameter of the loss, we assume only a bounded $k$-th moment. This assumption allows for unbounded, heavy-tailed stochastic gradient distributions, and can yield sharper excess risk bounds. Prior work characterized the minimax optimal rate for $\rho$-zero-concentrated DP SCO up to logarithmic factors in this setting, but the pure $\varepsilon$-DP case has remained open. We characterize the minimax optimal excess-risk rate for pure $\varepsilon$-DP heavy-tailed SCO up to logarithmic factors. Our algorithm achieves this rate in polynomial time with high probability. Moreover, it runs in deterministic polynomial time when the worst-case Lipschitz parameter is polynomially bounded. For important structured problem classes – including hinge/ReLU-type and absolute-value losses on Euclidean balls, ellipsoids, and polytopes – we achieve deterministic polynomial time even when the worst-case Lipschitz parameter is infinite. Our approach is based on a novel framework for privately optimizing Lipschitz extensions of the empirical loss. We complement our upper bound with a nearly matching high-probability lower bound.

This study introduces a hybrid deep learning model for intrusion detection in Industrial IoT (IIoT) systems, combining ResNet-1D, BiGRU, and Multi-Head Attention (MHA) for effective spatial-temporal feature extraction and attention-based feature weighting. To address class imbalance, SMOTE was applied during training on the EdgeHoTset dataset. The model achieved 98.71% accuracy, a loss of 0.0417%, and low inference latency (0.0001 sec /instance), demonstrating strong real-time capability. To assess generalizability, the model was also tested on the CICIoV2024 dataset, where it reached 99.99% accuracy and F1-score, with a loss of 0.0028, 0 % FPR, and 0.00014 sec/instance inference time. Across all metrics and datasets, the proposed model outperformed existing methods, confirming its robustness and effectiveness for real-time IoT intrusion detection.

We prove that no continuous, utility-preserving wrapper defense-a function $D: X\to X$ that preprocesses inputs before the model sees them-can make all outputs strictly safe for a language model with connected prompt space, and we characterize exactly where every such defense must fail. We establish three results under successively stronger hypotheses: boundary fixation-the defense must leave some threshold-level inputs unchanged; an $\epsilon$-robust constraint-under Lipschitz regularity, a positive-measure band around fixed boundary points remains near-threshold; and a persistent unsafe region under a transversality condition, a positive-measure subset of inputs remains strictly unsafe. These constitute a defense trilemma: continuity, utility preservation, and completeness cannot coexist. We prove parallel discrete results requiring no topology, and extend to multi-turn interactions, stochastic defenses, and capacity-parity settings. The results do not preclude training-time alignment, architectural changes, or defenses that sacrifice utility. The full theory is mechanically verified in Lean 4 and validated empirically on three LLMs.

CubeSats have revolutionized access to space by providing affordable and accessible platforms for research and education. However, their reliance on Commercial Off-The-Shelf (COTS) components and open-source software has introduced significant cybersecurity vulnerabilities. Ensuring the cybersecurity of CubeSats is vital as they play increasingly important roles in space missions. Traditional security measures, such as intrusion detection systems (IDS), are impractical for CubeSats due to resource constraints and unique operational environments. This paper provides an in-depth review of current cybersecurity practices for CubeSats, highlighting limitations and identifying gaps in existing methods. Additionally, it explores non-cyber anomaly detection techniques that offer insights into adaptable algorithms and deployment strategies suitable for CubeSat constraints. Open research problems are identified, including the need for resource-efficient intrusion detection mechanisms, evaluation of IDS solutions under realistic mission scenarios, development of autonomous response systems, and creation of cybersecurity frameworks. The addition of TinyML into CubeSat systems is explored as a promising solution to address these challenges, offering resource-efficient, real-time intrusion detection capabilities. Future research directions are proposed, such as integrating cybersecurity with health monitoring systems, and fostering collaboration between cybersecurity researchers and space domain experts.

LLM agents increasingly draft messages on behalf of users, yet users routinely overshare sensitive information and disagree on what counts as private. Existing systems support only suppression (omitting sensitive information) and generalization (replacing information with an abstraction), and are typically evaluated on single isolated messages, leaving both the strategy space and evaluation setting incomplete. We formalize privacy-preserving LLM communication as an Information Sufficiency (IS) task, introduce free-text pseudonymization as a third strategy that replaces sensitive attributes with functionally equivalent alternatives, and propose a conversational evaluation protocol that assesses strategies under realistic multi-turn follow-up pressure. Across 792 scenarios spanning three power-relation types (institutional, peer, intimate) and three sensitivity categories (discrimination risk, social cost, boundary), we evaluate seven frontier LLMs on privacy at two granularities, covertness, and utility. Pseudonymization yields the strongest privacy\textendash utility tradeoff overall, and single-message evaluation systematically underestimates leakage, with generalization losing up to 16.3 percentage points of privacy under follow-up.