Papers

Extracting MITRE ATT&CK techniques from cyber threat intelligence (CTI) reports is an open-set, multi-label problem requiring both high recall (not missing techniques) and high precision (not hallucinating unsupported ones). Existing rule-based, supervised, and LLM-based methods struggle to achieve both: rule-based and supervised approaches lack generalizability across diverse attack descriptions, while LLM-based approaches that couple candidate generation and validation within a single inference step suffer from limited recall and precision simultaneously. We propose CTI2TTP, which addresses this challenge through a diverge-then-converge design inspired by how human analysts work: first extracting broadly, then verifying rigorously. Reports are first decomposed into atomic behaviors, each anchored by a deterministic span localization stage to a specific evidence window in the source text. Candidate techniques are then proposed broadly for each behavior, and a convergent verification stage retains only candidates supported by both the localized evidence and the authoritative MITRE definition. We contribute two evaluation resources, a cleaned TRAM benchmark (TRAM-Clean) and a new annotated dataset (CTI2TTP-Bench), to address known annotation noise in existing benchmarks and elevate the task to document-level TTP extraction. On TRAM-Clean and CTI2TTP-Bench, CTI2TTP achieves 76.48% and 87.39% macro-F1 respectively, outperforming chain-of-thought prompting with the same backbone by 63.5% and 29.4%.

Network Intrusion Detection Systems (NIDS) are now increasingly leveraging Machine Learning (ML) techniques to detect malicious network activities. Numerous papers have scrutinized the security of ML-based NIDS (ML-NIDS) by testing them against various attacks involving adversarial perturbations. The findings were oftentimes worrying: by making imperceptible changes to a given input, powerful ML models would be bypassed. In this context, we took a step back and wondered: where (i.e., in what "space") have these perturbations been applied? We argue that real-world adversaries can apply adversarial perturbations only by operating on the hosts they can control – a concept which we define as _host-space perturbations_. To some, such an observation may seem trivial. And yet, through a systematic literature review (n=316), we found that prior work applied perturbations by manipulating pre-collected datapoints (e.g., a packet _captured by the router_, or a network flow _analysed by the ML-NIDS_). Such operations, while not impossible, may be outside the reach of an attacker who can only control some (unprivileged) hosts in a network. Hence, to demonstrate how to craft host-space perturbations and study some of their effects, we experimented on well-known benchmarks and a real-world network. We show that ML-NIDS that can detect the SSH-bruteforcing attempts launched via a given command string cannot detect any attempt launched by changing _a single character_ of such a string. We then examined how such a minuscule change in the "problem space" (i.e., the attacker's host) can lead to devastating effects on the "feature space". We derive lessons learned on how to practically assess host-space perturbations. Our stance is that the security of ML-NIDS should be re-assessed.

Membership inference attacks (MIAs) are popular methods for empirically assessing the leakage of sensitive information in the training data through models or statistics learned from the data. The MI vulnerability is often evaluated through a binary classifier that tries to predict whether a particular sample was in the training data. In order to evaluate the effectiveness of MIAs multiple shadow models are trained using random partitions of a larger dataset. After training the shadow models the MI vulnerability can be evaluated for all the samples for which we obtained shadow models. In order to evaluate the MI vulnerability reliably one needs a lot of shadow models which can be computationally infeasible. Therefore instead of reporting the actual sample level vulnerabilities aggregates over multiple samples are often reported in practice. We demonstrate two key weaknesses in typical MIA evaluation pipeline. First, we show that sampling the shadow datasets from a fixed superset leads to finite sample bias inflating the vulnerability estimates. Second, we show that evaluating the true positive rate (TPR) by concatenating MIA scores across multiple individuals, commonly used in the very low false positive rate (FPR) regime, is not calibrated across the per-sample FPRs. For both weaknesses we propose fixes that in the most simple approximate form do not incur any additional computation cost. We show that with additional computation one can further improve the reliability of the vulnerability estimation.

Semantic-level watermarking (SWM) improves robustness against text modifications by treating sentences as the basic unit. However, robustness to paragraph-level paraphrasing remains difficult because such attacks globally disrupt watermark signals by changing sentence order. In this work, we propose SAMark, a self-anchored watermarking framework that removes the dependency on sentence order by establishing a step-independent green region in semantic space. To improve detectability, we introduce a multi-channel hyperbolic scoring mechanism that amplifies watermark signals while suppressing noise from weakly aligned candidates. We further propose a diversity-aware filtering strategy that combines hard filtering with soft regularization, extending beyond simple n-gram repetition filters to address semantic redundancy. Experimental results show that SAMark achieves up to 90.2% TP@FP1% under typical paragraph-level paraphrasing attacks, outperforming the strongest prior baseline by more than 30% on average, while maintaining generation quality competitive with unwatermarked text and breaking the robustness-quality trade-off that limits prior methods. Our code will be released at [this URL](https://github.com/Z1zs/SAMark).

With the rapid development of mobile computing technology, massive amounts of spatial data are continuously generated from various mobile terminals and sensing devices, such as smartphones, connected vehicles, and drones. Performing efficient distributed statistical analysis on this data is crucial for real-time mobile computing applications. However, the constrained and dynamic nature of mobile environments exacerbates the privacy challenge: centralizing sensitive data for analysis risks severe privacy leaks, while existing privacy-preserving techniques often introduce excessive overhead or inaccuracies. In this paper, we design, implement, and evaluate the first system that supports efficient and privacy-preserving distribution statistics analysis for mobile spatial data. First, we propose eSpat-B, which leverages two non-colluding servers and a newly designed improved distributed point functions (DPF) with octree partitioning. Furthermore, considering the frequent updates of spatial data, we propose another more efficient scheme, eSpat+. The core idea of this scheme is to utilize a K-Dimensional tree for spatial partitioning, combine it with incremental DPF for performing statistics analysis, and design an efficient update algorithm. Security analysis demonstrates that our schemes effectively protect data privacy throughout the statistical process. Extensive experiments on real-world trajectory datasets demonstrate that the proposed schemes significantly outperform existing approaches, reducing computation overhead by up to 1.2x and communication overhead by up to 20x while maintaining 100% statistical accuracy.

Retrieval-Augmented Generation (RAG) empowers LLMs with external knowledge, making cross-institutional domain-specific knowledge base integration a highly promising deployment paradigm. Despite this potential, strict privacy regulations create severe "data silos" that obstruct such collaboration. Building federated RAG systems requires distributed inference, but the Transformer's self-attention mechanism fundamentally conflicts with this by mandating cross-node access to distributed Key-Value caches. To address this challenge, we present FedRAG, a high-throughput, privacy-preserving federated RAG framework. At its core is a novel Scrambled Distributed Attention protocol that utilizes numerically stable feature scrambling and token permutation. By dynamically delegating scrambled computations to collaborating nodes, our system successfully decouples attention execution from data localization without exposing plaintext. Crucially, our approach requires no specialized hardware or model retraining, circumventing the prohibitive latency and communication overheads of cryptographic solutions while robustly defending against intermediate state inversion attacks. Extensive evaluations demonstrate our framework preserves negligible (<0.1%) model utility degradation and achieves up to a 62$\times$ latency reduction over existing secure baselines, sustaining practical, human-reading throughput for cross-institutional knowledge synergy.

Mobile gaming apps increasingly rely on third-party Software Development Kits SDKs for advertising, analytics, attribution, and user engagement, potentially introducing privacy exposure beyond traditional permission based risks. Existing studies have largely focused on permissions or isolated tracking behaviors, providing only a partial understanding of privacy exposure in modern mobile ecosystems. This study presents a configuration aware empirical assessment of privacy exposure in Android mobile gaming apps by examining permissions, manifest level configurations, exported components, and SDK ecosystem complexity across children-oriented and general-audience games. A systematic static analysis was conducted on 41 widely deployed Android mobile gaming apps collected from the Google Play ecosystem. The analysis incorporated SDK categorisation and statistical evaluation using Spearman correlation, Mann Whitney U, and Chi square testing. The results revealed that privacy exposure is strongly associated with ecosystem-level architectural decisions rather than permission requests alone. Child-oriented games frequently demonstrated exposure conditions comparable to general-audience apps despite sometimes requesting fewer sensitive permissions. Furthermore, larger and more diverse SDK ecosystems were significantly associated with elevated privacy exposure levels, while advertising-oriented SDKs showed strong association with high exposure classifications. These findings highlight the limitations of permission-centric assessment approaches and emphasize the importance of configuration aware and ecosystem-aware privacy evaluation methodologies for modern mobile software systems.

Homomorphic quantum error correction aims to protect quantum data against both unauthorized access and environmental noise during server-based processing. We investigate the algebraic compatibility between quantum homomorphic encryption and quantum error correction, determining precise conditions under which encrypted encoded states remain inside the relevant code space during storage and computation. Our work establishes a necessary and sufficient criterion for an $[[n,1,d]]$ stabilizer code to remain compatible with the restricted transversal block-Pauli masking $U_{\rm enc}(a,b)=(X^aZ^b)^{\otimes n}$, stated explicitly for $[[n,1,d]]$ codes and extending directly to code-space preservation for $[[n,k,d]]$ codes. We verify this condition for standard examples (bit-flip and Shor codes, with the phase-flip repetition code following analogously), derive a practical criterion for Calderbank-Shor-Steane codes, and extend the analysis to three-dimensional color codes. A critical challenge emerges for non-Clifford gate implementation: the Shor code lacks a naive transversal $T$-gate implementation of the desired logical operation on encrypted encoded data. We present two routes around this obstruction. First, suitable triorthogonal codes admit transversal $T$-type logical implementations, up to Clifford corrections. Second, logical-gate masking gives code-space compatibility for arbitrary stabilizer codes, provided that suitable unitary representatives of the required logical gates are available. These results separate code-space compatibility from a full cryptographic security proof and provide explicit criteria for combining error correction with homomorphic processing in cloud quantum computing.

Security evaluations inherently depend on stable identifiers. Any finding, audit, or regulatory decision must remain attached to the specific artifact it pertains to. Continuously updated artificial intelligence systems violate this core assumption, with public model designations remaining static while underlying weights, prompts, retrieval mechanisms, misuse classifiers, inference settings, and serving infrastructures undergo unannounced modifications. Consequently, current evaluations frequently apply to superficial labels rather than identifiable and distinct systems. To resolve this, we propose referential security as a new paradigm for AI evaluation. The fundamental security question extends beyond whether a model is safe to whether subsequent parties can conclusively determine which system a specific safety claim addressed. This approach reframes model identity as an empirically verifiable property and separates referential stability from the substantive security claims it conditions. This framework brings tractability to three critical workflows that current practices handle poorly. Specifically, it enables reproducible evaluation, longitudinal audit validity, and cross-provider equivalence. By grounding these evaluations in verifiable artifacts, our approach ensures that safety audits and regulatory findings maintain their empirical utility across the operational lifecycle of dynamic systems.

We extend the closed-form privacy-subsidy result of Nakamura~(2026, arXiv:2605.15746) from the single-period Kyle model to continuous-time. A committed Bayesian automated market maker observes the aggregate order flow perturbed by an independent Brownian privacy channel of diffusion intensity $\sigma_\varepsilon$. Under the Markovian linear equilibrium, the price-impact coefficient is $\lambda = \sigma_v / \sqrt{\sigma_u^2 + \sigma_\varepsilon^2}$ – constant in time – and the cumulative expected transfer from the protocol's liquidity pool to traders over $[0,1]$ is $|\Pi_M| = \sigma_v \sigma_\varepsilon^2 / \sqrt{\sigma_u^2 + \sigma_\varepsilon^2}$. We then establish a structural correspondence between this cumulative privacy subsidy and Loss-Versus-Rebalancing (Milionis et al.~2022), identifying privacy-noise welfare as the order-flow observation analog of LVR's price observation gap. The result completes the continuous-time Kyle leg of the program of quantifying break-even fees for committed-AMM exchanges under privacy-aggregated information environments.

Extended Berkeley Packet Filter (eBPF) programs are kernel extensions used for networking, observability, and security enforcement in the Linux kernel. The in-kernel eBPF verifier checks low-level memory safety and termination on eBPF programs, but it does not enforce many higher-level source-level properties, such as initialization discipline, schema consistency, or error handling. We document nine classes of source-level bugs that compile, pass the kernel verifier, and can silently corrupt data, leak kernel memory to userspace, or yield incorrect enforcement outcomes. To harden such verifier-accepted buggy programs and support safe migration, we present Heimdall, an automated pipeline that uses large language models to translate legacy libbpf C programs to Aya Rust. Heimdall iteratively repairs compilation and kernel-verifier failures, rejects unsafe escape hatches in Rust-Aya with a static analysis safety engine, and proves per-program equivalence to the original via symbolic execution and Z3-based equivalence checking. Across 115 eBPF programs, Heimdall generates 109 formally proven-equivalent translations (94.8%). In the process, Heimdall identifies nine bugs in real-world eBPF programs and fixes all of them, two of which leak randomized kernel addresses to userspace and break KASLR. Eight of the nine have also been acknowledged and fixed upstream by the developers. Heimdall is the first system to automate memory-safe-language migration of production eBPF programs with per-program formal guarantees that the migration preserves observable behavior.

While Large Language Model-based Multi-Agent Systems (LLM-MAS) demonstrate remarkable capabilities in solving complex tasks by orchestrating specialized agents and external tools, the implicit trust in tool outputs creates a critical attack surface. Existing tool attacks are limited by domain specificity or fixed and static templates. To address these challenges, we propose Evo-Attacker, which formulates the tool attack as a self-evolving, memory-augmented reinforcement learning process. Evo-Attacker constructs a dynamic attack memory and employs deliberative reasoning to retrieve adversarial patterns and strategize modifying interventions at critical moments. Furthermore, we introduce Attack-Flow GRPO to optimize intermediate reasoning steps via terminal outcomes, addressing the long-horizon credit assignment challenge. Comprehensive experiments demonstrate that Evo-Attacker consistently outperforms baselines, highlighting its generalization and evolutionary capabilities and the urgent need for defensive tool safeguards.

KYA (Know Your Agents) is an open-source, framework-agnostic trust and governance layer for autonomous systems, composed of five primitives: (1) a four-gate inbound apply pipeline; (2) an only-tighten composition algebra over a three-channel multi-tenant hierarchy; (3) KYP (Know Your Principal), a schema-level unification of trust scoring across human users, AI agents, and service accounts; (4) auditable interaction-multiplier amplification over an AIVSS-shaped additive baseline; and (5) two-axis delegation attribution: a static premium for risky delegates and a runtime debit for actual delegate misbehavior in multi-agent fan-out. Together these span three pillars (trust, governance, and evidentiary assurance), making an autonomous system's actions authorized, policy-conforming, and post-hoc verifiable: where observability answers how long, how much, and what path, KYA answers was it authorized, did it conform, and can it be verified; it composes with observability rather than replacing it. It ships native adapters for 15+ agent frameworks. On a 4 by 9 cross-backend matrix all 36 cells pass; the pure-function scorer runs sub-millisecond at p99 and the system sustains ~ 1,800 ops/sec at 20 concurrent workers with HMAC chain integrity preserved end-to-end. KYA detects 89% of 1,200 adversarial probes from PyRIT and Garak, including the recently-published topology-guided multi-agent attack. The system is available under Apache 2.0 as the veldt-kya package on PyPI.

Heterogeneous Internet of Things (IoT) systems suffer from fragmentation across hardware architectures, networking stacks, and data serialization formats. Existing standards (such as MQTT, COAP, and DDS) rely on address-bound, imperative routing models that require hardcoded configurations and leave no flexibility for runtime schema translation. This paper presents TIP (The Intent Protocol), a decentralized, declarative network protocol. Instead of addressing specific physical endpoints, nodes submit abstract intents specifying desired capabilities, schemas, and Quality of Service (QoS) constraints. The TIP Engine resolves matching nodes using a hybrid discovery mechanism combining local multicast DNS (mDNS) with Kademlia Distributed Hash Tables (DHT). Selection is optimized via a multi-criteria scoring algorithm incorporating network latency, historical reputation, and contract compliance. Mismatched data representations are reconciled on-the-fly inside isolated WebAssembly (WASM) sandboxes compiled dynamically from TOML specifications. Security is enforced through Ed25519 signatures, X25519 key exchanges, and ChaCha20-Poly 1305 payload encryption. Evaluation of our reference implementation in Rust and C++ shows sub-millisecond translation overhead and robust resilience under industrial conditions.

Concept Bottleneck Models (CBMs) have emerged as a cornerstone approach for interpretable machine learning, providing human-understandable intermediate representations through explicit concept activations. However, this interpretability fundamentally introduces a critical, previously unexplored attack surface: the concept bottleneck layer itself. We present a comprehensive, systematic study of concept-level adversarial vulnerabilities in CBMs, revealing that targeted, minimal perturbations operating on input pixels can induce catastrophic misclassification by manipulating semantic representations. We develop a rigorous theoretical framework to quantify concept-space robustness, establishing novel metrics that expose the vulnerability landscape of these architectures. Our extensive analysis on the CUB-200-2011 dataset demonstrates that standard CBMs exhibit severe susceptibility to concept-level manipulation. To address this critical weakness, we introduce SPECTRA (Semantic Perturbation-based Concept Training for Robustness against Attacks), a principled stability regularization defense. SPECTRA effectively hardens the semantic representation space, increasing the minimal perturbation norm required for a successful attack from 0.46 to over 4,200, rendering targeted concept manipulation computationally prohibitive. Furthermore, SPECTRA preserves baseline classification accuracy to within 2.2%. By establishing concept-level attacks as a fundamentally distinct threat model, this work opens a new research frontier at the intersection of interpretable machine learning and adversarial robustness.

The rapid proliferation of Internet of Things (IoT) devices has created an urgent demand for adaptive, resource-efficient Intrusion Detection Systems (IDS) capable of handling dynamic and evolving cyber threats. This paper investigates AOC-IDS, a state-of-the-art autonomous online IDS published at IEEE INFOCOM 2024, which employs an Autoencoder (AE) with Cluster Repelling Contrastive (CRC) loss and an autonomous Gaussian-based decision module. We first successfully replicate AOC-IDS on the UNSW-NB15 benchmark, achieving 89.39% accuracy in close agreement with the published 89.19%. We then identify four key limitations: class imbalance, unreliable pseudo-label generation, limited generalization, and computational overhead for IoT deployment, and propose targeted improvements for each. Our XGBoost-BalSamp method achieves 95.45% accuracy on UNSW-NB15, a gain of 6.26% over the baseline. Our combined deep learning approach (PseudoFilter, MixupAug, and LiteAE) achieves a best-run accuracy of 90.88% (F1: 91.45%), surpassing the base paper while reducing model parameters by 55%.These results demonstrate that targeted improvements to AOC-IDS yield consistent accuracy gains while improving practical deployability on IoT edge devices.

Large language models are increasingly used as orchestrators of external tools via the Model Context Protocol (MCP), but MCP is built for software services with megabytes of memory and does not descend to the microcontrollers that dominate the long tail of physical devices. Recent work (IoT-MCP) ports MCP to edge gateways at 74 KB peak memory; this still excludes the smallest commodity MCUs and, critically, does not address the safety problem of giving an unreliable caller (an LLM that may hallucinate or be prompt-injected) direct control of physical hardware. We present the Device Context Protocol (DCP): a sub-50-byte typical frame (6-byte header + CBOR payload + optional 16-byte HMAC), a manifest schema in which capability scoping, range and type checks, dry-run evaluation, and units-as-types are protocol-layer primitives, and a host-side Bridge that rejects malformed or hallucinated calls before any byte reaches the device. Reference firmware measures 27.6 KB flash / 0.6 KB RAM on ESP32; the Python Bridge, ESP32 firmware, and a language-neutral conformance suite are MIT-licensed and public. An empirical study – 675 tool calls produced by five LLMs across four vendors (DeepSeek, Alibaba, Zhipu, MiniMax) against six categories of adversarial prompts, with the injection category instantiating AgentDojo's attack templates – shows DCP rejects 100% of capability-escalation attempts and 78% of prompt-injection attempts, versus 0–1% for Raw MCP and IoT-MCP, matching the expressiveness of a well-formed OpenAPI 3 schema at three orders of magnitude less firmware footprint. We position DCP as the missing layer between MCP (which is moving toward enterprise SaaS connectivity) and the physical devices it does not reach.

Safety alignment in large language models (LLMs) and multimodal large language models (MLLMs) is commonly assumed to operate as a near-binary threshold mechanism. We challenge this assumption by revealing that safety behavior is governed by an instability region where small perturbations induce stochastic refusal decisions rather than deterministic outcomes. We develop a multi-metric diagnostic framework combining external and internal signals to characterize this instability. Through systematic experiments, we identify a characteristic diagnostic signature: inputs in unstable regimes exhibit elevated output uncertainty yet decreased internal safety activation, a decoupling phenomenon that explains why detection-based defenses fail against sophisticated attacks. Building on this framework, we introduce Furina, a jailbreak attack that deliberately induces this signature through fragmented, scene-anchored prompts without model-specific optimization. Furina outperforms strong single-turn and multi-turn baselines on HarmBench and achieves competitive results on MM-SafetyBench, demonstrating that uncertainty amplification provides a principled and transferable mechanism for understanding safety vulnerabilities. Code is available at: https://github.com/0xCavaliers/Furina_Jailbreak.

The known stylistic biases in LLM judges, such as a preference for verbosity or specific sentence structures, present an underexplored security vulnerability. In this work, we introduce BITE (BIas exploraTion and Exploitation), a black-box adversarial framework that learns semantics-preserving edits to mislead an LLM judge and artificially inflate the scores it assigns. We cast the selection of stylistic edits as a contextual bandit problem and use a LinUCB policy to adaptively choose edits that maximize the judge's score without access to model parameters or gradients. Empirically, we test BITE across a diverse range of LLM judges and tasks, including both pointwise and pairwise comparisons on chatbot leaderboards and AI-reviewer benchmarks. BITE achieves an attack success rate exceeding 65% and raises scores by 1-2 points on a 9-point scale, all while preserving semantic equivalence. We further assess the attack's stealthiness, showing that BITE evades standard style-control methods and several detection baselines. Our findings expose a fundamental weakness in the LLM-as-a-judge paradigm and motivate robust, attack-aware evaluation. Our code is available at https://github.com/xianglinyang/llm-as-a-judge-attack.

LLM-driven agents are capable of selecting external tools to complete users' tasks. However, attackers could compromise such process, steering agents toward inappropriate/wrong tools and enabling malicious actions. Most existing attacks primarily manipulate the tool metadata, which is easily detectable by auditing and may lose effectiveness as modern agents increasingly adopt memory modules to refine tool selection policies through accumulated experience. This paper proposes MemMorph, the first attack that bias tool selection by poisoning the agent's long-term memory. Rather than explicitly dictating the tool invocation decision, MemMorph injects a small number of crafted records that are disguised as technical facts, incident reports, and operational policies. These poisoned records reshape the agent's contextual perception and decision-making process, leading it to autonomously infer and select the tool preferred by the attacker. Experiments across 3 benchmarks, 10 agent backbones, and 3 memory-module implementations show that MemMorph achieves up to 85.9% attack success rate with only three injected records, outperforming the strongest baseline by up to 25% while retaining potency under 3 representative defenses. Our findings expose long-term memory as a critical and under-explored attack surface in tool-augmented agents, urging the development of memory-level integrity safeguards.

Reentrancy attacks remain a persistent threat to decentralized applications (DApps), with malicious actors siphoning around 80M USD from the DApp ecosystem last year by exploiting EVM's inter-contract message-passing semantics. Existing research focuses primarily on detection, relying on known attack patterns, and fails to provide deployable solutions that eliminate the vulnerability. Traditional reentrancy guards are similarly limited, offering incomplete coverage across attack variations and lacking robustness against complex DApp interactions. In this paper, we introduce Sentinel, a novel proxy-based approach that mitigates reentrancy vulnerabilities in a type-agnostic way by integrating reentrancy logic directly into the proxy layer, intercepting all calls to the underlying implementation contract. Key features include a dual-mode operational system offering both a gas-optimized internal guard and a high-security external lock registry for cross-contract reentrancy prevention. The proxy also intelligently handles static calls, enabling safe view-function execution while protecting against Read-Only Reentrancy (ROR) attacks. Through rigorous evaluation on a dataset of 70 vulnerable smart contracts, Sentinel achieves 100% security coverage across four major reentrancy attack categories, outperforming existing solutions by over 40%

In this work, we study the interface of the Brazilian e-Voting Machine (BVM) in the context of electromagnetic side-channel threats commonly referred to as TEMPEST attacks. In a TEMPEST attack against video displays, an eavesdropper uses Software-Defined Radios (SDRs) to recover sensitive information by intercepting electromagnetic emanations generated during video signal transmission. We emulate the BVM using a VGA monitor by leveraging publicly available information disclosed by the electoral authority, including technical specifications, operational rules of the system, and the official BVM interface. Based on this setup, we investigate whether the BVM interface gives rise to a distinctive spectral signature observable through its unintended electromagnetic emissions. Our findings show that design characteristics relevant to a nationwide electoral process – such as high image contrast, minimal on-screen information, and the prohibition of other electronic devices within the polling station – result in a simple and highly distinctive spectral signature that can be observed even through a wall in our experiments. Although our experiments do not involve actual BVM hardware, the results raise concerns regarding the system's susceptibility to TEMPEST attacks and highlight the need for further research on protective countermeasures. In this context, our findings may support the design of automatic jammers capable of adaptively targeting compromising frequencies. To the best of our knowledge, this is the first study investigating TEMPEST attacks in the context of an electronic voting system officially adopted by a country.

Background: Fine-tuning is central to adapting pre-trained Large Language Models (LLMs) to downstream tasks, but its reliance on training data, parameter updates, and reusable components opens entry points for attackers. Threats have evolved from data poisoning and weight tampering to agent manipulation and interface exploitation, yet existing reviews lack a unified framework spanning the full fine-tuning lifecycle. Objective: This paper presents a systematic survey of LLM fine-tuning security and establishes a lifecycle-based framework for comparing attacks and defenses, complemented by unified empirical evaluation. Methods: We divide attack and defense mechanisms into three phases by intervention timing: pre-tuning, during-tuning, and post-tuning. Within each phase, strategies are reviewed and contrasted to expose their evolution and limitations. Representative methods are then evaluated under a unified model, hardware, and protocol setup, with cross-phase experiments pairing attacks and defenses from different phases. Results: Attack effectiveness is highly model-dependent and non-monotonic with scale: weight-editing attacks effective on earlier models lose impact on modern open-source LLMs; cross-lingual backdoor transfer, reported as near-perfect at larger scales, fails entirely on tested 1B-4B models; and purely benign samples can compromise safety alignment in instruction-tuned models. Single-phase defenses rarely generalize across phases, and defense effectiveness depends jointly on model architecture and alignment state. Conclusion: We identify key open problems (configuration-robust defense, cross-phase defense composition, and embedding-space attacks beyond behavioral assumptions) and propose concrete future research directions.

As machine-learning workloads move onto quantum hardware they inherit a trust problem classical AI deployments have largely solved and quantum ones have not: quantum machine learning (QML) is already sold as a cloud service, and a customer cannot check that the circuit they paid for is actually the circuit that ran. We show the gap is exploitable. A malicious or cost-cutting provider can substitute the declared quantum channel with one that reproduces every accuracy test the customer performs, passes the single-observable checks used in practice, and still differs from the declared computation at the channel level. We call this a channel-substitution attack and instantiate it in three deployments: audit evasion, service downgrade, and result forgery in reproducibility review. QML-PipeGuard detects the attack at runtime from measurement statistics the customer already collects. Its guarantee is two-sided, stated in the distance a local verifier measures: any substitution shifting a single-qubit output marginal by $\delta$ in trace norm forces a contract deviation of at least $\delta/\sqrt{3}$, the constant tight and independent of qubit count; the guarantee is relative to the chosen observable family and reference-input set. For local-unitary substitutions the adversary's blind set is characterised exactly, as a commutation condition, which gives the defender a rule for choosing challenge inputs that shrinks it. A finite-shot budget rule bounds false accepts and false alarms at once, one tolerance separates attack from the benign drift observed here, and every decision enters a hash-anchored audit trail. On an IBM Heron r2 processor a substituted quantum-kernel intrusion-detection pipeline evades the weak contract and our verifier halts it, while the honest channel stays within tolerance.

Remote Direct Memory Access (RDMA) is a key enabler of high-performance systems, offering low latency, high throughput, and reduced CPU overhead by allowing direct memory-to-memory transfers between machines. However, its design bypasses traditional CPU-mediated security mechanisms, introducing critical vulnerabilities in untrusted environments. This work explores the integration of RDMA and AES-128 encryption to secure data transfers without compromising performance. We implement encryption directly within the data plane of a programmable Tofino switch using the P4 programming language. By offloading encryption from the CPU to the switch, our design preserves RDMA's performance benefits while addressing its security shortcomings. Experimental results show that the system achieves throughput of 0.37 Gbps for 16-byte packets, 0.76 Gbps for 32-byte packets, 1.83 Gbps for 64-byte packets, and 1.9 Gbps for 128-byte packets. These findings demonstrate the feasibility of secure, high-throughput RDMA communication using programmable network hardware.

Memory-backed agents need provenance that can survive leaked or migrated snapshots, where logs, visible outputs, and trusted metadata may be absent. We propose MemMark, a state-evolution attribution watermark that embeds an owner-controlled signal into latent memory-write decisions. At each internal LLM call, MemMark samples among admissible candidates using keyed, distribution-preserving selection, and records cryptographic commitments with signed session anchors and reveal evidence. This makes attribution depend on reproducible backend behavior rather than mutable provenance fields. Across A-Mem and Graphiti on LoCoMo, with three LLM backbones, MemMark preserves memory utility: Overall F1 retains 99.6% of the unwatermarked baseline, while BLEU-1 changes by +0.2%. It also provides usable carrier capacity, with 1.16, 1.14, and 1.26 bits of mean entropy for update-target, link-target, and semantic-realization decisions. In the snapshot-only R3 setting, MemMark recovers the full 40-bit payload from final snapshots, while wrong-key verification remains near chance. Under nine memory-lifecycle attacks, verification distinguishes tampering, evidence deletion, and partial payload recovery. These results show that robust snapshot-only attribution is feasible for long-term agent memory without surviving traces, trusted metadata, or utility-degrading.

The advent of digital technologies has revolutionized traditional power distribution networks, transforming them into smart grids that are more reliable, efficient, and sustainable. Despite these advancements, electricity theft remains a significant threat to the effective operation of large electrical networks. To address this issue, we propose EnThM, a lightweight and communication-efficient scheme for real-time mitigation of power theft in smart grid systems. Our approach uses the hierarchical structure of the smart grid infrastructure to verify the authenticity of the metering data at multiple levels of the power distribution network. Our work focuses primarily on issues related to cryptographic security. The verification process involves statistically modeling the cumulative averages of the power usage data and applying rule-based checks on the aggregated power consumption at each level, while accounting for seasonal and daily consumption variations. The proposed method has been tested on benchmark consumption data, yielding high accuracy, efficient implementation, and real-time applicability.

Penetration testing is essential to securing modern web infrastructures, yet traditional manual methods struggle to keep pace with their scale and complexity. Large Language Models (LLMs) offer new opportunities for automating these tasks, but existing approaches face two persistent challenges: hallucination of technical entities and insufficient long-term contextual memory. To address these issues, we present APT-Agent, a fully automated LLM-driven penetration testing framework that systematically orchestrates reconnaissance, exploitation, and exfiltration. APT-Agent introduces a hybrid rectification module to recover hallucinated commands and a command-specific memory architecture to preserve operational context across multi-step attack sequences. We evaluate our APT-Agent on Metasploitable 2 against seven vulnerable services spanning web, database, and network protocols. APT-Agent achieves an 84.29% end-to-end exploitation success rate, compared to 48.57% (Script Kiddie) and 18.57% (PentestGPT) under matched conditions. By reducing cognitive burden and minimizing reliance on human intervention, APT-Agent represents a step toward scalable, reliable, and cognitively efficient automation for penetration testing.

LLM agents increasingly combine long-term memory for personalization with tool calling for taking actions. We study a failure of this combination: personality-driven preferences stored in memory (e.g., cost-consciousness, impatience, risk tolerance) silently shifting tool-call parameters in contexts where they do not apply. We call this memory-induced tool-drift and stress-test it with MEMDRIFT, an iteratively refined benchmark of 105 scenarios spanning five bias dimensions and seven professional domains. Measured as excess parameter degradation over a neutral-memory baseline, biased memories in MEMDRIFT induce drift in all eight models we evaluate, raising degradation rates to 0.38–0.83 versus 0.05–0.12 for neutral memories. The drift persists under three widely used memory frameworks. A vulnerability audit of 6,062 tools across 288 MCP servers demonstrates that real-world tools are also susceptible to this failure mode. Mechanistically, we find that biased memories act as implicit steering vectors along the same direction as explicit behavioral instructions and draw attention away from task-relevant context. Neither prompt-based defenses nor relevance filters offer a complete solution, motivating defenses at the intersection of memory management and tool calling.

Machine learning based malware detectors become obsolete over time due to concept drift in benign and malware applications. Recent methods rely on fully labeled data and use hierarchical contrastive loss (HCL) with active learning to improve robustness against drift by exploiting semantic structure in malware representations. However, obtaining labeled data in the security domain is difficult. Under partially labeled settings, HCL suffers significant performance degradation in detecting unseen malware, especially on datasets such as BODMAS where strong semantic structure may not exist. In this paper, we propose SEED, a semantic-structure-agnostic method for malware detection under limited supervision. SEED combines a tailored binary cross-entropy objective with semi-supervised continual learning and active learning. For partially labeled seen tasks, unlabeled samples are projected into a representation space constructed from previously seen data using singular value decomposition, and paired with suitable labeled samples to encourage representation consistency. For unseen tasks with fully unlabeled data, uncertainty is quantified using cosine distance in representation space, and the most uncertain samples are selected for analyst labeling. We evaluate SEED on both Windows and Android malware datasets. Using only 20% labeled data on seen tasks, SEED achieves average AUT improvements of 40% on BODMAS and 14% on AndroZoo for unseen malware detection compared to HCL* (the semi-supervised adaptation of HCL), while remaining competitive on APIGraph. Finally, we introduce a delayed buffer update strategy to reduce label noise propagation during replay and improve learning stability.