Papers

This paper presents and characterizes a spectrum of previously unreported behaviours we term Constraint-Evasive Fabrication (CEF): when an LLM agent operates under irreconcilable constraints (where no response can simultaneously satisfy all active rules) it spontaneously fabricates plausible external obstacles and presents them as a fact. At the extreme end of this spectrum lies Constraint-Evasive Thanatosis (CET); the limit case where, rather than inventing a plausible excuse, the model simulates a full system crash to make the user disengage entirely. We first observed CET in an uncontrolled deployment test, where a GPT-4o banking agent fabricated Python-style exception traces (complete with memory addresses) to feign a system failure when threatened by a user. In subsequent controlled experiments, the model independently invented audit restrictions, microservice architectures, error codes, and service timeouts, none present in its prompt. Reproduction attempts across pressure levels and attacker personas yielded CEF consistently but with substantial variation in form, onset, and severity: the phenomenon is robust but stochastic. Critically, injecting ground-truth data mid-conversation did not restore honest behaviour once fabrication had taken hold (the model ignored correct information and continued confabulating) suggesting CEF is self-reinforcing rather than a knowledge gap. We show that (1) standard enterprise guardrails routinely create CEF-enabling conditions in production, (2) current RLHF procedures suppress but cannot eliminate CEF, and (3) existing safety benchmarks do not test for this failure mode. Our results highlight the need for irreconcilable-constraint benchmarks, CEF-aware training procedures, and deployment-time detection methods before constrained agents become further entrenched in high-stakes domains.

This paper presents a structured analysis of security challenges in long-horizon agentic AI systems. The study reviews existing threats, evaluation approaches, attack propagation mechanisms, and security frameworks. A taxonomy of security threats and a framework for analyzing attack propagation are proposed to support future research in agentic AI security

Verifier reliability is not portable across tasks. A verifier-guided self-DPO pipeline with genuine held-out gains on MathVista (+9.6 points on self-training data, +8.0 held out) can be harmful on MMMU. The failure is invisible from the target-task self-training signal: over six learner-verifier configurations, MMMU self-training performance still improves by +3.52 points while held-out performance drops by 1.42 points. We call this silent negative transfer: a verifier validated as useful before can still show signatures of successful self-training after its induced update stops transferring to unseen data. Downstream failure alone cannot diagnose this: poor generalization may mean the verifier induces a wrong learning direction, or that the correct direction itself does not generalize beyond the self-training distribution. We separate these effects via gradient fidelity F, the alignment between verifier-induced and correct training directions, and gradient transferability T, the alignment between correct and held-out directions, giving a conservative Safe-Transfer Margin: positive first-order held-out alignment is certified when arccos(F) + arccos(T) < pi/2. We then introduce Asymmetric Acceptance Gating (AAG), a precision-first filter selecting preference pairs by the verifier's absolute confidence in the response receiving the positive update. On MMMU, AAG raises fidelity from 0.29 to 0.42. In a same-cell analysis, the correct direction stays positively aligned with the held-out direction (T = 0.31), yet raw verification rotates the induced update to negative held-out alignment (-0.13); AAG restores positive alignment (+0.22). Across all six MMMU configurations, AAG beats raw verifier-guided training. Verifier reliability thus means not that it worked before, but that it induces the right update on the task at hand and that this update stays useful beyond the data that generated it.

Blockchain interoperability enables independent blockchain systems to communicate and exchange assets across heterogeneous networks. However, the lack of comprehensive security mechanisms remains a critical weakness – one that attackers have already exploited to cause hundreds of millions of dollars in asset losses. This paper presents a systematic identification and classification of security threats facing interoperable blockchain systems, along with corresponding countermeasures for each. We organize threats into five categories: (1) core blockchain attacks, (2) network attacks, (3) interoperability-specific attacks, (4) social engineering, and (5) code vulnerabilities, with particular attention to smart contract weaknesses. For each identified threat, we analyze its attack surface and propose effective defensive strategies. The resulting taxonomy provides a structured foundation for designing and evaluating secure blockchain interoperability solutions.

Browser automation frameworks are essential tools for security and privacy research on the web, yet bot detection scripts increasingly probe their artifacts, threatening measurement validity as automated browsers may be blocked or served different content. Prior work measures detection deployment, while we measure blocking-induced sample loss. Through a literature survey of top-tier security, privacy, and web measurement venues, we find that 83% of papers omit any discussion of bot detection blocking. To address this gap, we conduct a measurement study of 10,000 websites across four browser configurations (40K page visits in total) to quantify detection prevalence and employed techniques. Using custom instrumentation to detect when sites probe for automation, we develop a taxonomy of bot detection techniques and measure how often they appear in practice. Chromium headless encounters a 15% soft block rate compared to 7% for other configurations. Across all conditions, 82% of blocks are attributable to bot detection (59% vendor-confirmed, 23% inferred from condition-dependent blocking), predominantly by providers with integrated bot detection such as Cloudflare (37% block rate) and Akamai (26%). A header spoofing experiment establishes that 75% of Chromium-headless-only blocks are caused by header-level signals alone, yet JavaScript-based environment probing is more extensive than current blocking rates suggest. These findings demonstrate that bot detection creates systematic, provider-correlated sample loss that the web measurement community neither measures nor reports. The downstream effect on specific measurement outcomes remains future work.

LLM-based guardrails have emerged as a highly effective defense against prompt injection and jailbreak attacks in autonomous agents. However, we reveal that the very reasoning and task-following capabilities enabling this protection introduce a novel vulnerability: attackers can inject crafted data to trap the guardrail in extended reasoning loops, effectuating a systematic denial-of-service (DoS) attack. To systematically expose this threat, we design a beam-search optimization framework that crafts natural-language payloads to maximize guardrail reasoning length, utilizing an LLM proposer guided by a strategy bank. Based on the observation of guardrail's schema-following nature, we also provide another attack framework driven by mechanism-aware structural mutations with less computational load. The attack efficacy is systematically evaluated in two parts. First, in standalone evaluations, the attack generalizes across diverse guardrail architectures, safety templates, and agent benchmarks. Payloads optimized on a single open-source surrogate successfully transfer to eight leading model backbones (e.g., Claude, GPT, Gemini, DeepSeek, and Qwen), achieving a 13–63$\times$ token amplification. Second, in end-to-end real-world agent deployments (web, desktop, code, and multi-agent systems), the attack reveals up to a 148$\times$ latency amplification. We show that a single poisoned document can saturate shared guardrail infrastructures, effectively starving co-located agents and paralyzing the entire system. By uncovering this availability flaw, our work underscores the urgent need to develop cost-bounded, reasoning-robust guardrails.

Internet of Medical Things (IoMT) devices operate under strict resource constraints while handling highly sensitive health data, making security and privacy critical concerns. Federated learning (FL) further complicates this landscape, as model updates exchanged during training may unintentionally expose private medical information. Emerging quantum computing capabilities threaten the long-term viability of conventional lightweight cryptographic mechanisms, motivating the integration of Post-Quantum Cryptography (PQC) into IoMT systems. This article discusses key enabling technologies for quantum-resilient IoMT, including post-quantum key establishment, lightweight encryption, and edge-native orchestration. We propose a scalable Kubernetes-based framework that integrates PQC into FL-enabled IoMT environments and validate it on a Raspberry Pi testbed. Results demonstrate that distributed cryptographic processing significantly reduces latency compared to sequential designs while maintaining feasible resource overhead. The primary contribution of this work lies in the design and validation of a secure orchestration and communication framework for FL-enabled IoMT systems. We conclude by outlining future directions toward energy-aware architectures, intelligent security optimization, and resilient next-generation Intelligent Internet of Medical Things (IIoMT) ecosystems.

Most TinyML hardware accelerators focus on supporting Quantized Neural Networks (QNNs) to meet stringent constraints on power consumption and size. Despite this, the security aspects of quantization within TinyML hardware remain largely unexplored. Although previous studies indicate that QNNs demonstrate similar or enhanced robustness when compared to full-precision Deep Neural Networks (DNNs) against typical evasion attacks, no attack strategies tailored specifically for TinyML hardware have been proposed yet. This paper addresses this shortfall by demonstrating how a two-step attack pipeline can surpass the current state-of-the-art in the QNN context and shows the need for more hardware-aware security research.

In contemporary IoT edge devices with real-time requirements, security is primarily enforced through design-time parameters associated with security tasks, leading to mechanisms that operate in an opportunistic manner. As a result, security checks are often performed as secondary operations. This approach can result in systems where no security tasks are executed due to high utilization by other tasks. An alternative approach taken in prior work is to add security mechanisms to every task in the system, resulting in substantially lower performance than that of a system with no security. These approaches have resulted in an all-or-nothing scenario for edge device security, motivating numerous studies on the safety-security trade-off in real-time cyber-physical systems (RT-CPS). This study introduces an analytical framework – REPOSE – for evaluating the security feasibility of real-time control systems at runtime. REPOSE is developed for weakly-hard real-time control systems that facilitate a “bounded trade-off” between safety and security. In contrast to imposing additional (pessimistic) design-time overhead as considered in some real-time security literature, REPOSE performs security operations in both proactive and reactive manners based on the task's current behavior. Our evaluations show that REPOSE can effectively add security operations to RT-CPS with a feasibility overhead of $0.06\%$ at $80\%$ utilization, compared to a $ 29\%$ overhead observed in systems with hard constraints. Through a case study of a classic control system, we also demonstrate that REPOSE provides a robust framework to analyze and calculate the safety-security tradeoff.

Frontier AI systems are increasingly capable of cybersecurity tasks, including codebase inspection, vulnerability detection, and exploitation. However, evaluating their offensive capabilities remains constrained by limited access to open, reproducible, multi-host cyber ranges. Existing public benchmarks capture isolated skills such as CTF solving, vulnerability reproduction, and exploit generation, but often abstract away realistic intrusion workflows: discovering exposed services, gaining a foothold, collecting internal information, and expanding compromise across hosts. This gap makes it difficult to observe emerging risks early, because frontier AI systems are rarely evaluated under realistic attack conditions. We introduce AgentCyberRange, the first open, multi-range infrastructure for measuring autonomous cyber attack capability in realistic cyber ranges. It combines 110 vulnerabilities across 15 real web applications and 8 enterprise-like cyber ranges with 156 internal hosts, plus Cage, a toolchain for execution, orchestration, result collection, and verification. The benchmark covers two core stages: web exploitation, where agents explore exposed applications and validate vulnerabilities, and post exploitation, where agents turn an initial foothold into broader internal compromise. We evaluate six frontier AI systems under matched prompts and budgets. GPT-5.5 with Codex performs best, solving 16.1% of web exploitation tasks and 31.7% of post-exploitation tasks; with more concrete hints, these rates increase to 33.0% and 46.3%. We also observe out-of-benchmark findings, including unknown vulnerabilities in popular projects, and payload mutation that bypasses host defenses. These results show that open cyber-range evaluation is necessary for observing emerging offensive capabilities under realistic and reproducible conditions.

Secure software engineering in practice is a multi-stage workflow involving vulnerability analysis, remediation, and fix verification. However, current LLM-based software security approaches often focus on isolated tasks such as detection or patch generation, with limited attention to agentic architectures reflecting industrial workflow. This creates a gap between existing LLM-based vulnerability-handling methods and real-world practices. In this paper, we study a role-based agentic workflow for vulnerability analysis and mitigation consisting of Planner, Analyzer, Fixer, and Verifier roles. To explore the effect of static analysis tool, the analyzer agent was integrated with the CodeQL in one of the workflows. The models used include nemotron-cascade-2:30b, qwen3-coder-next, and gpt-oss:120b. Our evaluation uses 25 real-world C/C++ vulnerabilities. The study reports 44% vulnerability detection accuracy comparable to GPT 5.5 and 19% fix accuracy. We also list implications from this study in context of software security practitioners.

Android applications (apps) developers increasingly rely on code obfuscation techniques to hinder reverse engineering and protect intellectual property. However, obfuscation also reduces the effectiveness of static analysis and vulnerability detection tools, creating challenges for Android security analysis. Existing approaches for detecting obfuscation in Android apps predominantly rely on handcrafted heuristics, engineered features, or task-specific learning pipelines, which may struggle to generalize across evolving obfuscation strategies. This paper presents a large-scale empirical study investigating the capability of Large Language Models (LLMs) to detect obfuscation in Android apps through semantic reasoning. Our study evaluates whether off-the-shelf LLMs can identify obfuscated code without relying on handcrafted rules, predefined signatures, or dedicated model training. The empirical evaluation is conducted on both a controlled benchmark containing an app obfuscated with multiple techniques and a real-world dataset of Android apps collected from Google Play. The study further examines the impact of prompt design, model selection, and decision thresholds across several open-weight and proprietary LLMs. Finally, the analysis compares LLM-based reasoning with existing SAST-based obfuscation-detection approaches and discusses the broader implications and limitations of applying LLMs to Android security analysis.

Large language models (LLMs) are increasingly deployed in privacy-sensitive domains, where users must balance the risk of data exposure through external APIs against the high computational cost of local deployment. Split learning has therefore emerged as a promising paradigm for LLM fine-tuning and inference under limited local resources. However, it introduces new privacy risks. Prior work primarily studies leakage of private input prompts, typically via inversion attacks on intermediate representations, while the potential for sensitive information leakage through generative response outputs remains largely unexplored. In this work, we unveil novel vulnerabilities of Split-LLM by presenting Patched Model Inversion with Dual-Sided Initialization (PIDI), a two-stage attack that simultaneously targets both private input prompts and output responses in Split-LLM settings. It combines dual-sided initialization with a patched inversion strategy to tackle long sequences, substantially outperforming prior inversion methods. To counter threats from both sides, we further propose the Adapter-based DualGuard with Mutual Information Defense (ADMI), which integrates an adapter-based local warmup strategy and mutual information regularization to provide a strong empirical privacy protection with minimal impact on task performance. Extensive experiments across diverse tasks and models demonstrate that ADMI effectively defends against PIDI and other state-of-the-art inversion attacks. Our code is publicly available at https://github.com/FLAIR-THU/VFLAIR-LLM.

The rapid digitalization of the Sudanese financial sector has precipitated a surge in Mobile Banking Applications (MBAs); however, this growth has frequently outpaced rigorous security auditing. This study provides a comprehensive technical audit of the four most widely used Sudanese MBAs( Bankak, Fawry, Okash, and Sahil )collectively serving a user base of over 1.6 million. Utilizing Static Application Security Testing (SAST) via the Mobile Security Framework (MobSF) and Quixxi, the applications were evaluated against the OWASP Mobile Application Security Verification Standard (MASVS). Findings were mapped to Common Weakness Enumeration (CWE) identifiers to identify systemic vulnerabilities. Analysis revealed critical disparities in security posture. Bankak, the market leader, exhibited the highest risk profile (12 vulnerabilities), including a critical absence of SSL certificate pinning and unsafe TrustManager implementations, rendering it highly susceptible to Man-in-the-Middle (MitM) attacks. While Fawry demonstrated relative maturity (7 vulnerabilities), a universal failure was observed across all four applications regarding secure random number generation (CWE-330), potentially compromising session token integrity. Additionally, Bankak and Okash were found to utilize deprecated cryptographic algorithms (MD5/SHA-1). Notably, all applications successfully disabled ADB backups, yet 100% retained verbose debugging symbols in production APKs, significantly lowering the barrier for reverse engineering. This research addresses a critical gap in the national fintech ecosystem by providing actionable technical recommendations for developers and a strategic roadmap for implementing "security-by-design" principles across the sector.

Fuzz drivers are essential components of greybox fuzzing, as they encapsulate target interfaces, define test spaces, and largely determine fuzzing effectiveness. Existing fuzz drivers typically rely on crash-based oracles for security testing, overlooking library functionality and limiting bug detection capability. In this paper, we present the first study on metamorphic-based fuzz oracle enhancement (MFOE), which augments existing fuzz drivers with metamorphic-based oracles derived from metamorphic relations (MRs). Since constructing and integrating such oracles requires substantial domain knowledge, automating MFOE is challenging. To address this challenge, we propose MetaFOE, an LLM-based framework that automatically generates and integrates metamorphic-based oracles. We evaluate MetaFOE on OSS-Fuzz drivers using three modern LLMs and five prompt strategies. MetaFOE generates 3,475 MRs, of which 77.3% are applicable, and implements 12,351 meta drivers, with 6,228 being valid. After three hours of fuzzing, the valid meta drivers improve edge coverage by an average of 18.7% and trigger 1,528 unique crashes. Our results demonstrate both the effectiveness of metamorphic-based oracle enhancement and the feasibility of using LLMs to automate MFOE, providing valuable insights for advancing greybox fuzzing.

Large language model (LLM) agents increasingly extend their capabilities at runtime by loading Agent Skills, which pair natural-language specifications (SKILL.md) with executable scripts and resources. Because a skill's behavior relies on both natural-language instructions and executable code, assessing its safety requires cross-modal reasoning, creating a new language-and-code attack surface. Attackers can present a benign workflow in SKILL.md while embedding implicit directives that steer the agent to exfiltrate sensitive files, even if the scripts appear harmless. This attack surface remains understudied; prior work treats skills merely as prompt-injection vectors or static code artifacts, leaving attacks emerging from cross-modal interactions largely unmeasured. In our evaluation, open-source and commercial skill scanners detect only 2%-8% and 9%-17% of such attacks, respectively. To address this gap, we introduce SkillMutator, the first benchmark for install-time detection of language-and-code cross-modal attacks on Agent Skills. It emulates an adversarial mutation process across 13 attack categories, iteratively refining malicious skills using scanner feedback to make injected behaviors indistinguishable from legitimate workflows. We further propose a four-phase reasoning-trajectory distillation framework to distill frontier-teacher traces into smaller open-weight models. This produces a locally deployable scanner avoiding third-party data exposure and excessive API costs. On the strongest SkillMutator subset (n=76), our distilled model (Qwen2.5-Coder-7B-Instruct) improves detection from 17.1% to 88.2%, surpassing GPT-4o-mini (23.7%) and GPT-5.4-mini (79.0%), and reaching frontier-level GPT-5.4 (86.8%). These results show practical defense against cross-modal attacks is feasible without relying on costly frontier models.

In hierarchical organizations, authenticating data from multiple users can be complex and resource-intensive. Hierarchical Identity-Based Signature with Designated Aggregator (HIBS-DA) provides an efficient solution by allowing users at different levels to generate signatures that can be combined into a single, compact signature. We first introduce the HIBS-DA framework and present the {\em{first}} lattice-based construction of HIBS-DA. Our scheme allows users at different hierarchical levels to generate individual signatures that can be aggregated into a single, compact signature, reducing communication and verification costs. The proposed construction is secure, correct, and resistant to forgery, making it suitable for large-scale environments such as universities, corporations, and government agencies.

The modern software supply chain, taking Node Package Manager (npm) dependency network for example, relies heavily on shared open-source dependencies. While this promotes rapid development, it introduces systemic vulnerabilities as well. Concerning this potential risk, we analyze the npm dependency network by modeling 53,481 packages and 78,520 dependency edges, and classify the network as a scale-free topology. Thus, we demonstrate its inherent vulnerability to targeted attacks on high-degree hubs. To mitigate this, we propose and evaluate a dual-pronged defense strategy consisting of Centrality-Based Node-Hardening and Dependency Weight Warning system. Moreover, by simulating the network under various attack scenarios, we prove that applying strict security protocols to just the top 1% of nodes, combined with pruning 30% of structurally trivial edges, prevents catastrophic network collapse and neutralizes cascading malware infections. The source code can be found at https://github.com/5tarWhee1/Centrality-Based-Protection-Strategy-for-Supply-Chain-Security-in-npm-Dependency-Network.

Agentic browsers integrate autonomous AI agents into web browsers, enabling users to accomplish web tasks through natural-language instructions. The same-origin policy (SOP) is a fundamental browser security mechanism that prevents unauthorized automated cross-origin data flows induced by scripts. However, whether SOP remains effective in agentic browsers is an open question that has not been systematically studied. In this work, we bridge this gap. We first observe that an agentic browser can itself serve as an automated channel for cross-origin data flows, potentially leading to SOP violations. To investigate this phenomenon, we construct SOPBench, a benchmark for evaluating SOP violations in agentic browsers. Our evaluation shows that existing agentic browsers frequently violate SOP, both in benign settings and under attacks. To address this problem, we propose SOPGuard, an SOP enforcement mechanism tailored to agentic browsers. We implement SOPGuard in BrowserOS, an open-source agentic browser. Extensive evaluations demonstrate that SOPGuard effectively enforces SOP while preserving utility and incurring only a small runtime overhead. Our code and data are available at https://github.com/wxl-lxw/BrowserOS-SOPGuard.

In recent years, the Institute of Electrical and Electronics Engineers (IEEE) and the European Telecommunications Standards Institute (ETSI) have developed a series of security communication standards for vehicular communications. These standards include mechanisms such as the Security Credential Management System (SCMS) and Butterfly Key Expansion (BKE) to protect vehicle privacy. However, these standards are mainly based on the Elliptic-Curve Cryptography (ECC), which may be vulnerable to attacks from quantum computing in the future. In response to this potential risk, this study proposes a hybrid certificate that combines the ECC with Post-Quantum Cryptography (PQC). This approach enables infrastructure systems to be built on cryptographic foundations that are more resilient to quantum-based attacks. Furthermore, this study presents a generalized pseudonym scheme that is compatible with various cryptographic algorithms for generating pseudonym certificates. This design aims to eliminate the possibility of inferring any correlation between the public key in a pseudonym certificate and that in an enrollment certificate. This study also conducts a comprehensive performance evaluation of the RSA, ECC, and PQC algorithms, particularly those standardized by the National Institute of Standards and Technology (NIST). The comparison considers factors such as message length and computation time. Based on the findings, this study recommends suitable pseudonym schemes that adopt hybrid certificates for secure and efficient use in vehicular communications.

LLM-based Agents are becoming increasingly capable and widely deployed, creating growing incentives for adversarial misuse in the real-world. A key emerging threat is Decomposition Attacks \cite{glukhov2024breach, jones2024adversaries} in which a harmful task is broken into simpler, benign subtasks that evade safety mechanisms when executed separately but cumulatively fulfill the malicious intent. Although recent benchmarks assess agent safety in multi-turn and multi-tool-use settings, they do not explicitly capture this form of decompositional misuse and may not represent realistic adversarial execution flows. To this end, we introduce DeCompBench, a benchmark designed specifically to evaluate agentic safety under decomposition attacks. DeCompBench is created with a decomposition-by-design principle using a graphical framework and enables harmful task decomposition into individually benign and executable subtasks with realistic workflows. Our experiments using a custom decomposer show that state-of-the-art agents exhibit high refusal rates on monolithic harmful tasks, but significantly lower refusal rates on their decomposed variants, while often inadvertently fulfilling the adversarial objectives. These findings underscore the need for safety evaluations against decomposition attacks and corresponding defenses. Our dataset is publicly available and can be found at https://huggingface.co/datasets/decompositionbench/DeCompBench.

Privacy-preserving exchange designs price on a coarsened view of order flow. We show that a market maker committed to informationally efficient (posterior-mean) pricing on a signal strictly coarser than the flow it settles necessarily cedes a closed-form welfare transfer to traders – the privacy subsidy – and that no rule restricted to the coarse signal is simultaneously efficient and zero-profit against the settled flow. We establish this impossibility for a general coarsening, then characterise the subsidy in closed form across three canonical microstructure models: single-period Kyle with Gaussian flow noise, Glosten-Milgrom with a binary direction channel, and continuous-time Kyle-Back with a Brownian channel. The subsidy obeys a structural correspondence with Loss-Versus-Rebalancing, both welfare rates factorising as a squared noise driver times a committed-object factor. Gross of fees the subsidy is a pure transfer recovered by a break-even fee; once levied, that fee distorts volume, and the resulting deadweight – under an explicit allocative value of trade – is fourth-order in the noise scale while the gross subsidy is second-order, so privacy is welfare-neutral to leading order with a strictly smaller irrecoverable loss. Endogenising the privacy level, a protocol trading a differential-privacy benefit against this deadweight chooses an interior noise scale in closed form; doing so leaves the half-revealing product of price impact and informed intensity intact while unpinning the volatility-elasticity of price impact from its textbook value of one.

This paper develops a finite certificate calculus for ambient release systems, staged probabilistic environments in which a protected coordinate is not observed directly but can remain statistically readable through visible roles, timing, repeated movement, bounded attention, linked rooms, and post-release state. The security notion, choric masking, requires the trace law induced by a protected locus to lie inside or near the convex hull of admissible cover traces under the tests available to a specified audience. For finite horizons, trace laws form polytopes, audiences induce measurement operators, and masking becomes intersection in the projected measurement space. Exposure is certified by separating hyperplanes, kernel obstructions, hypothesis-testing bounds, Fano-type localization lower bounds, and support separation in downstream rooms. The calculus distinguishes trace residue from carrier localization, full-trace exposure from attention-filtered exposure, first-room masking from delayed post-release exposure, and unresolved system pressure from carrier hazard. It proves measurement-polytope equivalence for exact and approximate masks, dual separation certificates, data-processing laws for attention lenses, aperture identities for gaze-thinned observation, lower bounds for mandatory unique gestures, composition rules for linked releases, and a repeated-room debt theorem showing how unresolved pressure can broaden selection and shift cost onto cover populations without localizing the source. The result is a finite, checkable language for auditing privacy, unlinkability, side-channel leakage, and accountability in public-facing release systems.

Modern software supply chains have evolved into vast, heterogeneous networks where transparency - the granular understanding of all software components - is now a critical security requirement. While Software Bills of Materials (SBOMs) have emerged as the primary mechanism for this transparency, current industry practices rely on a metadata-centric paradigm that assumes an artifact is defined solely by its package manager declarations. We posit that this assumption is fundamentally flawed, creating a systemic visibility gap we define as Software Dark Matter (SDM). SDM represents the set of security-critical files present in an artifact's filesystem that are unaccounted for by its associated metadata. We implement a reference tool, DARKFILES, and use it to analyze four ecosystems of disjoint nature: DockerHub, Maven Central, plugin/extension marketplaces (Jenkins plugins and OpenVSX), and a real-world enterprise environment. Our research makes the following contributions: we introduce a general-purpose metric for artifact fidelity calculating SDM as the ratio of untracked files per total file count. We introduce Packaging Lag, a phenomenon where official metadata remains out-of-date across multiple versions before catching up to an artifact's actual content. We demonstrate that SDM exposes vulnerable software invisible to SBOM-driven pipelines both by cross-referencing untracked packages against known CVE databases and through the direct discovery of three confirmed high-severity CVEs, showing that SDM is highly correlated with sensitive information including secrets and cryptographic keys.

Active Motor Noise Cancellation (AMNC) is a noise-reduction feature shipped in commercial fused deposition modeling (FDM) 3D printers. Because it suppresses the acoustic emissions that side-channel attacks exploit, it has security-relevant side effects, though we find no evidence it was designed as a security control. We present a duration-controlled evaluation of side-channel leakage on AMNC-equipped hardware, using a public dataset of 144 synchronized acoustic and vibration recordings from two Bambu Lab printers across 12 object classes. Spectral analysis confirms suppression is measurably active: the motor-resonance band rises only 4.92 dB above its background-relative baseline during printing. Leakage nonetheless survives it. Acoustic classification is at chance for 30-second windows (11.11%, permutation p = 0.188) but reaches 27.08% on a duration-clean 60-second window and 40.28% under distributed sampling, against an 8.33% baseline: the channel is suppressed within short windows, not eliminated. The dominant discriminator, however, is print duration itself, which classifies at 63.89% (95% CI [55.78, 71.28]) from recording length alone and is validated against sliced G-code print time at r = 0.907. Vibration carries genuine geometry information independent of duration: with the observation window equalized by truncation, classification reaches 45.83% against a 25% baseline (permutation p = 0.023). It nonetheless adds no measurable information beyond duration in paired comparison. The leak is architecture-specific, present on the core-XY device (36.11%) and indistinguishable from chance on the bed-slinger (13.89%), and a consumer handset recovers as much as a mounted accelerometer (25.00% vs 26.39%). Noise cancellation raises the observation time an acoustic attacker requires without eliminating the channel, and leaves print duration entirely untouched.

We present a Bayesian calibration layer for slopsquat detectors – those that flag hallucinated package imports in code produced by large language models (LLMs). Where existing pipelines emit binary decisions (flag / do-not-flag), our layer emits a Beta-posterior probability per detection, derived from a 3-category epistemic taxonomy that explicitly classifies each prior as empirically calibrated, constructively argued, or engineering-judgement-traced. Beyond the primary 200/404 registry channel, the calibrated layer exploits PyPI metadata signals – package age, release count, author descriptor, summary – to surface registered-but-suspicious packages that a binary registry detector misses, which is the realistic post-LLM-emission attacker regime. The resulting risk-aware primitive is directly consumable by downstream CI gates and supports principled threshold decisions across detection rules. We evaluate the calibration on a merged corpus of 1,734 Python snippets – a stratified 189-prompt BigCodeBench slice plus a 100-prompt niche-library stress-test set, generated across a six-model panel spanning four cloud models (Claude-Sonnet-4.6, Mistral-Large, DeepSeek-v4-pro, DeepSeek-R1) and two local open-weight code models (Mistral Codestral, Meta CodeLlama). Against a re-implemented binary baseline inspired by Mahmud et al. – which shares its registry oracle with our ground truth and therefore serves as a degenerate upper bound rather than a genuine competitor – the calibrated layer reproduces the strict-registry detections and introduces well-calibrated additional flags on the metadata channel. We assess detector asymmetry with a McNemar paired test and calibration with both a flagged-subset Expected Calibration Error and a strictly proper full-corpus Brier score.

The intersection of crypto x AI is spawning papers, products, online posts, and companies. All the surrounding buzz, though, obscures what exactly has been done, what the opportunities and challenges are, and what open questions deserve attention. This survey paper asks what AI can do for blockchain-based technologies (broadly construed as "crypto") (crypto x AI), and vice versa (AI x crypto). We systematize existing work, summarize key takeaways, highlight open research questions, and offer a perspective on pervasive industry misconceptions, concluding that AI and crypto are still in the very early stages of meaningful integration.

Hardware Security at Willamette is a Willamette University affiliated research group studying the hardware-software interface of security critical services. Within our program, we noticed many researchers spent considerable development time learning to understand and manually parse traces-of-execution of hardware designs which are used to identifying whether vulnerabilities or weaknesses arise at the hardware, software, or interface level. We propose the "RTL-Arrow" framework, a framework to compile performant binaries which bridge the hardware/data divide. We translate the outputs of simulated hardware execution, as "value change dumps" into modern data science workflows as cloud-ready "dataframes", to standardize program verification across the hardware and software levels. We describe our approach, its benefits, and lessons learned from the process of packaging and distributing these libraries for our security research program.

Security validation is an important yet challenging part of the hardware design process, yet, by convention, validation engineers are tasked with defining the threat model, specifying the relevant security properties, detecting any violations of those properties, and assessing the consequences to system security, each of which is manually intensive and may introduce errors. The combined technologies of information flow tracking and specification mining represent an automated approach to property generation and validation, but prior work on information flow tracking on RTL trace data was limited to find cases under which information flowed between registers, without reproducing full paths to capture how sensitive information propagates through a design. With the introduction of new technologies accelerating hardware analysis, we develop a novel approach for constructing information flow paths from register transfer level (RTL) trace data.

Autonomous network-security response systems promise to reduce Security Operations Centre (SOC) reaction latency, but reward-only multi-agent reinforcement learning (MARL) can improve security reward while remaining non-deployable. We present a safety-contract graph MARL framework and instantiate it as ACD$^3$-GAT (Adaptive Constrained Counterfactual Decisioning with a Graph Attention Network encoder), an architecture that separates simulator observations from reusable operational budgets, constrained optimization, graph state encoding, and counterfactual action screening. We evaluate the method in CAGE Challenge 4, where agents operate under budgets for Mean Time to Recover (MTTR), false-positive response, and firewall change-management disruption. Across the benchmark, every unconstrained method violates the SOC downtime budget in 100% of evaluated episodes, with mean downtime proxy costs of 311-430 against a budget of 50. This complements prior CAGE Challenge 4 findings by showing that reward-only learning lacks operational discipline. Constrained MAPPO-GAT (C-MAPPO-GAT) isolates Lagrangian operational-cost control and budget-aware screening, while ACD$^3$-GAT adds budget context, CVaR tail-risk estimation, opponent-belief state, and Graph Counterfactual Risk Propagation (G-CRP). The replicated comparison includes three 200-episode seeds for IPPO, MAPPO-GAT, C-MAPPO-GAT, and ACD$^3$-GAT. C-MAPPO-GAT reduces downtime violation from 100% to 0.3% and mean downtime cost from 355.4 to 15.5 relative to MAPPO-GAT. ACD$^3$-GAT reduces mean downtime cost to 48.2 with a 13.8% violation rate, placing it on the safety-contract frontier rather than at the most conservative compliance point. Topology-seed and coupled adaptive Red-process stress tests preserve this contrast and show lower worst adaptive degradation for safety-constrained policies than reward-only MAPPO-GAT.