Papers

The generalization capability of deepfake detectors is critical for real-world use. Data augmentation via synthetic fake face generation effectively enhances generalization, yet current SoTA methods rely on fixed strategies-raising a key question: Is a single static augmentation sufficient, or does the diversity of forgery features demand dynamic approaches? We argue existing methods overlook the evolving complexity of real-world forgeries (e.g., facial warping, expression manipulation), which fixed policies cannot fully simulate. To address this, we propose CRDA (Curriculum Reinforcement-Learning Data Augmentation), a novel framework guiding detectors to progressively master multi-domain forgery features from simple to complex. CRDA synthesizes augmented samples via a configurable pool of forgery operations and dynamically generates adversarial samples tailored to the detector's current learning state. Central to our approach is integrating reinforcement learning (RL) and causal inference. An RL agent dynamically selects augmentation actions based on detector performance to efficiently explore the vast augmentation space, adapting to increasingly challenging forgeries. Simultaneously, the agent introduces action space variations to generate heterogeneous forgery patterns, guided by causal inference to mitigate spurious correlations-suppressing task-irrelevant biases and focusing on causally invariant features. This integration ensures robust generalization by decoupling synthetic augmentation patterns from the model's learned representations. Extensive experiments show our method significantly improves detector generalizability, outperforming SOTA methods across multiple cross-domain datasets.

Large-scale Video Foundation Models (VFMs) has significantly advanced various video-related tasks, either through task-specific models or Multi-modal Large Language Models (MLLMs). However, the open accessibility of VFMs also introduces critical security risks, as adversaries can exploit full knowledge of the VFMs to launch potent attacks. This paper investigates a novel and practical adversarial threat scenario: attacking downstream models or MLLMs fine-tuned from open-source VFMs, without requiring access to the victim task, training data, model query, and architecture. In contrast to conventional transfer-based attacks that rely on task-aligned surrogate models, we demonstrate that adversarial vulnerabilities can be exploited directly from the VFMs. To this end, we propose the Transferable Video Attack (TVA), a temporal-aware adversarial attack method that leverages the temporal representation dynamics of VFMs to craft effective perturbations. TVA integrates a bidirectional contrastive learning mechanism to maximize the discrepancy between the clean and adversarial features, and introduces a temporal consistency loss that exploits motion cues to enhance the sequential impact of perturbations. TVA avoids the need to train expensive surrogate models or access to domain-specific data, thereby offering a more practical and efficient attack strategy. Extensive experiments across 24 video-related tasks demonstrate the efficacy of TVA against downstream models and MLLMs, revealing a previously underexplored security vulnerability in the deployment of video models.

Deep neural networks have recently achieved notable progress in 3D point cloud recognition, yet their vulnerability to adversarial perturbations poses critical security challenges in practical deployments. Conventional defense mechanisms struggle to address the evolving landscape of multifaceted attack patterns. Through systematic analysis of existing defenses, we identify that their unsatisfactory performance primarily originates from an entangled feature space, where adversarial attacks can be performed easily. To this end, we present 3D-ANC, a novel approach that capitalizes on the Neural Collapse (NC) mechanism to orchestrate discriminative feature learning. In particular, NC depicts where last-layer features and classifier weights jointly evolve into a simplex equiangular tight frame (ETF) arrangement, establishing maximally separable class prototypes. However, leveraging this advantage in 3D recognition confronts two substantial challenges: (1) prevalent class imbalance in point cloud datasets, and (2) complex geometric similarities between object categories. To tackle these obstacles, our solution combines an ETF-aligned classification module with an adaptive training framework consisting of representation-balanced learning (RBL) and dynamic feature direction loss (FDL). 3D-ANC seamlessly empowers existing models to develop disentangled feature spaces despite the complexity in 3D data distribution. Comprehensive evaluations state that 3D-ANC significantly improves the robustness of models with various structures on two datasets. For instance, DGCNN's classification accuracy is elevated from 27.2% to 80.9% on ModelNet40 – a 53.7% absolute gain that surpasses leading baselines by 34.0%.

As large language models (LLMs) become increasingly capable, concerns over the unauthorized use of copyrighted and licensed content in their training data have grown, especially in the context of code. Open-source code, often protected by open source licenses (e.g, GPL), poses legal and ethical challenges when used in pretraining. Detecting whether specific code samples were included in LLM training data is thus critical for transparency, accountability, and copyright compliance. We propose SynPrune, a syntax-pruned membership inference attack method tailored for code. Unlike prior MIA approaches that treat code as plain text, SynPrune leverages the structured and rule-governed nature of programming languages. Specifically, it identifies and excludes consequent tokens that are syntactically required and not reflective of authorship, from attribution when computing membership scores. Experimental results show that SynPrune consistently outperforms the state-of-the-arts. Our method is also robust across varying function lengths and syntax categories.

To prevent misinformation and social issues arising from trustworthy-looking content generated by LLMs, it is crucial to develop efficient and reliable methods for identifying the source of texts. Previous approaches have demonstrated exceptional performance in detecting texts fully generated by LLMs. However, these methods struggle when confronting more advanced LLM output or text with adversarial multi-task machine revision, especially in the black-box setting, where the generating model is unknown. To address this challenge, grounded in the hypothesis that human writing possesses distinctive stylistic patterns, we propose Human Language Preference Detection (HLPD). HLPD employs a reward-based alignment process, Human Language Preference Optimization (HLPO), to shift the scoring model's token distribution toward human-like writing, making the model more sensitive to human writing, therefore enhancing the identification of machine-revised text. We test HLPD in an adversarial multi-task evaluation framework that leverages a five-dimensional prompt generator and multiple advanced LLMs to create diverse revision scenarios. When detecting texts revised by GPT-series models, HLPD achieves a 15.11% relative improvement in AUROC over ImBD, surpassing Fast-DetectGPT by 45.56%. When evaluated on texts generated by advanced LLMs, HLPD achieves the highest average AUROC, exceeding ImBD by 5.53% and Fast-DetectGPT by 34.14%. Code will be made available at https://github.com/dfq2021/HLPD.

Steinke (2025) recently asked the following intriguing open question: Can we solve the differentially private selection problem with nearly-optimal error by only (adaptively) invoking Gaussian mechanism on low-sensitivity queries? We resolve this question positively. In particular, for a candidate set $\mathcal{Y}$, we achieve error guarantee of $\tilde{O}(\log |\mathcal{Y}|)$, which is within a factor of $(\log \log |\mathcal{Y}|)^{O(1)}$ of the exponential mechanism (McSherry and Talwar, 2007). This improves on Steinke's mechanism which achieves an error of $O(\log^{3/2} |\mathcal{Y}|)$.

Ensuring compliance with Information Flow Security (IFS) is known to be challenging, especially for concurrent systems with large codebases such as multicore operating system (OS) kernels. Refinement, which verifies that an implementation preserves certain properties of a more abstract specification, is promising for tackling such challenges. However, in terms of refinement-based verification of security properties, existing techniques are still restricted to sequential systems or lack the expressiveness needed to capture complex security policies for concurrent systems. In this work, we present a generalized security-preserving refinement technique, particularly for verifying the IFS of concurrent systems governed by potentially complex security policies. We formalize the IFS properties for concurrent systems and present a refinement-based compositional approach to prove that the generalized security properties (e.g., intransitive noninterference) are preserved between implementation and abstraction. The key intuition enabling such reasoning, compared to previous refinement work, is to establish a step-mapping relation between the implementation and the abstraction, which is sufficient to ensure that every paired step (in the abstraction and the implementation, respectively) is either permitted or prohibited by the security policy. We apply our approach to verify two non-trivial case studies against a collection of security policies. Our proofs are fully mechanized in Isabelle/HOL, during which we identified that two covert channels previously reported in the ARINC 653 single-core standard also exist in the ARINC 653 multicore standard. We subsequently proved the correctness of the revised mechanism, showcasing the effectiveness of our approach.

Safety alignment instills in Large Language Models (LLMs) a critical capacity to refuse malicious requests. Prior works have modeled this refusal mechanism as a single linear direction in the activation space. We posit that this is an oversimplification that conflates two functionally distinct neural processes: the detection of harm and the execution of a refusal. In this work, we deconstruct this single representation into a Harm Detection Direction and a Refusal Execution Direction. Leveraging this fine-grained model, we introduce Differentiated Bi-Directional Intervention (DBDI), a new white-box framework that precisely neutralizes the safety alignment at critical layer. DBDI applies adaptive projection nullification to the refusal execution direction while suppressing the harm detection direction via direct steering. Extensive experiments demonstrate that DBDI outperforms prominent jailbreaking methods, achieving up to a 97.88% attack success rate on models such as Llama-2. By providing a more granular and mechanistic framework, our work offers a new direction for the in-depth understanding of LLM safety alignment.

As Federated Learning (FL) becomes more widespread, there is growing interest in its decentralized variants. Decentralized FL leverages the benefits of fast and energy-efficient device-to-device communications to obviate the need for a central server. However, this opens the door to new security vulnerabilities as well. While FL security has been a popular research topic, the role of adversarial node placement in decentralized FL remains largely unexplored. This paper addresses this gap by evaluating the impact of various coordinated adversarial node placement strategies on decentralized FL's model training performance. We adapt two threads of placement strategies to this context: maximum span-based algorithms, and network centrality-based approaches. Building on them, we propose a novel attack strategy, MaxSpAN-FL, which is a hybrid between these paradigms that adjusts node placement probabilistically based on network topology characteristics. Numerical experiments demonstrate that our attack consistently induces the largest degradation in decentralized FL models compared with baseline schemes across various network configurations and numbers of coordinating adversaries. We also provide theoretical support for why eigenvector centrality-based attacks are suboptimal in decentralized FL. Overall, our findings provide valuable insights into the vulnerabilities of decentralized FL systems, setting the stage for future research aimed at developing more secure and robust decentralized FL frameworks.

Precise and sound call graph construction is crucial for many software security mechanisms. Unfortunately, traditional static pointer analysis techniques used to generate application call graphs suffer from imprecision. These techniques are agnostic to the application's architecture and are designed for broad applicability. To mitigate this precision problem, we propose PhaseSeed, a novel technique that improves the accuracy of pointer analysis for split-phase applications, which have distinct initialization and processing phases. PhaseSeed analyzes the initialization phase dynamically, collecting the points-to relationships established at runtime. At the end of the initialization phase, it then seeds this information to a static analysis stage that performs pointer analysis for all code that stays in scope during the processing phase, improving precision. Our observations show that, given the same runtime configuration options, the points-to relationships established during the initialization phase remain constant across multiple runs. Therefore, PhaseSeed is sound with respect to a given initial configuration. We apply PhaseSeed to three security mechanisms: control flow integrity (CFI), software debloating, and system call filtering. PhaseSeed provides up to 92.6% precision improvement for CFI compared to static call graph construction techniques, and filters nine additional security-critical system calls when used to generate Seccomp profiles.

Low-altitude wireless networks (LAWNs) have emerged as a viable solution for maritime communications. In these maritime LAWNs, unmanned aerial vehicles (UAVs) serve as practical low-altitude platforms for wireless communications due to their flexibility and ease of deployment. However, the open and clear UAV communication channels make maritime LAWNs vulnerable to eavesdropping attacks. Existing security approaches often assume eavesdroppers follow predefined trajectories, which fails to capture the dynamic movement patterns of eavesdroppers in realistic maritime environments. To address this challenge, we consider a low-altitude maritime communication system that employs intelligent jamming to counter dynamic eavesdroppers with uncertain positioning to enhance the physical layer security. Since such a system requires balancing the conflicting performance metrics of the secrecy rate and energy consumption of UAVs, we formulate a secure and energy-efficient maritime communication multi-objective optimization problem (SEMCMOP). To solve this dynamic and long-term optimization problem, we first reformulate it as a partially observable Markov decision process (POMDP). We then propose a novel soft actor-critic with conditional variational autoencoder (SAC-CVAE) algorithm, which is a deep reinforcement learning algorithm improved by generative artificial intelligence. Specifically, the SAC-CVAE algorithm employs advantage-conditioned latent representations to disentangle and optimize policies, while enhancing computational efficiency by reducing the state space dimension. Simulation results demonstrate that our proposed intelligent jamming approach achieves secure and energy-efficient maritime communications.

With the widespread application of large language models (LLMs) in various fields, the security challenges they face have become increasingly prominent, especially the issue of jailbreak. These attacks induce the model to generate erroneous or uncontrolled outputs through crafted inputs, threatening the generality and security of the model. Although existing defense methods have shown some effectiveness, they often struggle to strike a balance between model generality and security. Excessive defense may limit the normal use of the model, while insufficient defense may lead to security vulnerabilities. In response to this problem, we propose a Knowledge Graph Defense Framework (KG-DF). Specifically, because of its structured knowledge representation and semantic association capabilities, Knowledge Graph(KG) can be searched by associating input content with safe knowledge in the knowledge base, thus identifying potentially harmful intentions and providing safe reasoning paths. However, traditional KG methods encounter significant challenges in keyword extraction, particularly when confronted with diverse and evolving attack strategies. To address this issue, we introduce an extensible semantic parsing module, whose core task is to transform the input query into a set of structured and secure concept representations, thereby enhancing the relevance of the matching process. Experimental results show that our framework enhances defense performance against various jailbreak attack methods, while also improving the response quality of the LLM in general QA scenarios by incorporating domain-general knowledge.

Secure data hiding remains a fundamental challenge in digital communication, requiring a careful balance between computational efficiency and perceptual transparency. The balance between security and performance is increasingly fragile with the emergence of generative AI systems capable of autonomously generating and optimising sophisticated cryptanalysis and steganalysis algorithms, thereby accelerating the exposure of vulnerabilities in conventional data-hiding schemes. This work introduces SteganoSNN, a neuromorphic steganographic framework that exploits spiking neural networks (SNNs) to achieve secure, low-power, and high-capacity multimedia data hiding. Digitised audio samples are converted into spike trains using leaky integrate-and-fire (LIF) neurons, encrypted via a modulo-based mapping scheme, and embedded into the least significant bits of RGBA image channels using a dithering mechanism to minimise perceptual distortion. Implemented in Python using NEST and realised on a PYNQ-Z2 FPGA, SteganoSNN attains real-time operation with an embedding capacity of 8 bits per pixel. Experimental evaluations on the DIV2K 2017 dataset demonstrate image fidelity between 40.4 dB and 41.35 dB in PSNR and SSIM values consistently above 0.97, surpassing SteganoGAN in computational efficiency and robustness. SteganoSNN establishes a foundation for neuromorphic steganography, enabling secure, energy-efficient communication for Edge-AI, IoT, and biomedical applications.

Network steganography and covert communication channels have been studied extensively in the past. However, prior works offer minimal practical use for their proposed techniques and are limited to specific use cases and network protocols. In this paper, we show that covert channels in networking have a much greater potential for practical secret communication than what has been discussed before. We present a covert channel framework, CYPRESS, that creates a reliable hidden communication channel by mounting packets from secret network entities on regular packets that flow through the network, effectively transmitting a separate network traffic without generating new packets for it. CYPRESS establishes a consolidated decentralized framework in which different covert channels for various protocols are defined with their custom handler code that are plugged into the system and updated on-demand to evade detection. CYPRESS then chooses at run-time how and in what order the covert channels should be used for fragmentation and hidden transmission of data. We can reach up to 1.6MB/s of secret bandwidth in a network of ten users connected to the Internet. We demonstrate the robustness and reliability of our approach in secret communication through various security-sensitive real-world experiments. Our evaluations show that network protocols provide a notable opportunity for unconventional storage and hidden transmission of data to bypass different types of security measures and to hide the source of various cyber attacks.

Small language models (SLMs) are increasingly deployed on edge devices, making their safety alignment crucial yet challenging. Current shallow alignment methods that rely on direct refusal of malicious queries fail to provide robust protection, particularly against adversarial jailbreaks. While deliberative safety reasoning alignment offers deeper alignment for defending against sophisticated attacks, effectively implanting such reasoning capability in SLMs with limited capabilities remains an open challenge. Moreover, safety reasoning incurs significant computational overhead as models apply reasoning to nearly all queries, making it impractical for resource-constrained edge deployment scenarios that demand rapid responses. We propose EASE, a novel framework that enables practical and Efficient safety Alignment for Small languagE models. Our approach first identifies the optimal safety reasoning teacher that can effectively distill safety reasoning capabilities to SLMs. We then align models to selectively activate safety reasoning for dangerous adversarial jailbreak queries while providing direct responses to straightforward malicious queries and general helpful tasks. This selective mechanism enables small models to maintain robust safety guarantees against sophisticated attacks while preserving computational efficiency for benign interactions. Experimental results demonstrate that EASE reduces jailbreak attack success rates by up to 17% compared to shallow alignment methods while reducing inference overhead by up to 90% compared to deliberative safety reasoning alignment, making it practical for SLMs real-world edge deployments.

LockBit has evolved from an obscure Ransomware-as-a-Service newcomer in 2019 to the most prolific ransomware franchise of 2024. Leveraging a recently leaked MySQL dump of the gang's management panel, this study offers an end-to-end reconstruction of LockBit's technical, behavioral, and financial apparatus. We recall the family's version timeline and map its tactics, techniques, and procedures to MITRE ATT&CK, highlighting the incremental hardening that distinguishes LockBit 3.0 from its predecessors. We then analyze 51 negotiation chat logs using natural-language embeddings and clustering to infer a canonical interaction playbook, revealing recurrent rhetorical stages that underpin the double-extortion strategy. Finally, we trace 19 Bitcoin addresses related to ransom payment chains, revealing two distinct patterns based on different laundering phases. In both cases, a small portion of the ransom is immediately split into long-lived addresses (presumably retained by the group as profit and to finance further operations) while the remainder is ultimately aggregated into two high-volume addresses before likely being sent to the affiliate. These two collector addresses appear to belong to distinct exchanges, each processing over 200k BTC. The combined evidence portrays LockBit as a tightly integrated criminal service whose resilience rests on rapid code iteration, script-driven social engineering, and industrial-scale cash-out pipelines.

Safety evaluation of large language models (LLMs) increasingly relies on LLM-as-a-judge pipelines, but strong judges can still be expensive to use at scale. We study whether structured multi-agent debate can improve judge reliability while keeping backbone size and cost modest. To do so, we introduce HAJailBench, a human-annotated jailbreak benchmark with 11,100 labeled interactions spanning diverse attack methods and target models, and we pair it with a Multi-Agent Judge framework in which critic, defender, and judge agents debate under a shared safety rubric. On HAJailBench, the framework improves over matched small-model prompt baselines and prior multi-agent judges, while remaining more economical than GPT-4o under the evaluated pricing snapshot. Ablation results further show that a small number of debate rounds is sufficient to capture most of the gain. Together, these results support structured, value-aligned debate as a practical design for scalable LLM safety evaluation.

ROI selective encryption, as an efficient privacy protection technique, encrypts only the key regions in the video, thereby ensuring security while minimizing the impact on coding efficiency. However, existing ROI-based video encryption methods suffer from insufficient flexibility and lack of a unified evaluation system. To address these issues, we propose a visual perception-based tunable framework and evaluation benchmark for H.265/HEVC ROI encryption. Our scheme introduces three key contributions: 1) A ROI region recognition module based on visual perception network is proposed to accurately identify the ROI region in videos. 2) A three-level tunable encryption strategy is implemented while balancing security and real-time performance. 3) A unified ROI encryption evaluation benchmark is developed to provide a standardized quantitative platform for subsequent research. This triple strategy provides new solution and significant unified performance evaluation methods for ROI selective encryption field. Experimental results indicate that the proposed benchmark can comprehensively measure the performance of the ROI selective encryption. Compared to existing ROI encryption algorithms, our proposed enhanced and advanced level encryption exhibit superior performance in multiple performance metrics. In general, the proposed framework effectively meets the privacy protection requirements in H.265/HEVC and provides a reliable solution for secure and efficient processing of sensitive video content.

Large Language Models (LLMs) are widely adopted, but their high training cost leads many developers to fine-tune existing open-source models. While most adhere to open-source licenses, some falsely claim original training despite clear derivation from public models, raising pressing concerns about intellectual property protection and the need to verify model provenance. In this paper, we propose GhostSpec, a lightweight yet effective method for verifying LLM lineage without access to training data or modification of model behavior. Our approach constructs compact and robust fingerprints by applying singular value decomposition (SVD) to invariant products of internal attention weight matrices. Unlike watermarking or output-based methods, GhostSpec is fully data-free, non-invasive, and computationally efficient. Extensive experiments show it is robust to fine-tuning, pruning, expansion, and adversarial transformations, reliably tracing lineage with minimal overhead. By offering a practical solution for model verification, our method contributes to intellectual property protection and fosters a transparent, trustworthy LLM ecosystem. Our code is available at https://github.com/DX0369/GhostSpec.

At CRYPTO 2019, Gohr pioneered neural cryptanalysis by introducing differential-based neural distinguishers to attack Speck32/64, establishing a novel paradigm combining deep learning with differential cryptanalysis.Since then, constructing neural distinguishers has become a significant approach to achieving the deep learning-based cryptanalysis for block ciphers.This paper advances rotational-XOR (RX) attacks through neural networks, focusing on optimizing distinguishers and presenting key-recovery attacks for the lightweight block ciphers Simon32/64 and Simeck32/64.In particular, we first construct the fundamental data formats specially designed for training RX-neural distinguishers by refining the existing data formats for differential-neural distinguishers. Based on these data formats, we systematically identify optimal RX-differences with Hamming weights 1 and 2 that develop high-accuracy RX-neural distinguishers. Then, through innovative application of the bit sensitivity test, we achieve significant compression of data format without sacrificing the distinguisher accuracy. This optimization enables us to add more multi-ciphertext pairs into the data formats, further strengthening the performance of RX-neural distinguishers. As an application, we obtain 14- and 17-round RX-neural distinguishers for Simon32/64 and Simeck32/64, which improves the previous ones by 3 and 2 rounds, respectively.In addition, we propose two novel techniques, key bit sensitivity test and the joint wrong key response, to tackle the challenge of applying Bayesian's key-recovery strategy to the target cipher that adopts nonlinear key schedule in the related-key setting without considering of weak-key space. By this, we can straightforwardly mount a 17-round key-recovery attack on Simeck32/64 based on the improved 16-round RX-nerual distinguisher. To the best of our knowledge, the presented RX-neural......

This position paper argues that setting the privacy budget in differential privacy should not be viewed as an important limitation of differential privacy compared to alternative methods for privacy-preserving machine learning. The so-called problem of interpreting the privacy budget is often presented as a major hindrance to the wider adoption of differential privacy in real-world deployments and is sometimes used to promote alternative mitigation techniques for data protection. We believe this misleads decision-makers into choosing unsafe methods. We argue that the difficulty in interpreting privacy budgets does not stem from the definition of differential privacy itself, but from the intrinsic difficulty of estimating privacy risks in context, a challenge that any rigorous method for privacy risk assessment face. Moreover, we claim that any sound method for estimating privacy risks should, given the current state of research, be expressible within the differential privacy framework or justify why it cannot.

Software security testing, particularly when enhanced with deep learning models, has become a powerful approach for improving software quality, enabling faster detection of known flaws in source code. However, many approaches miss post-fix latent vulnerabilities that remain even after patches typically due to incomplete fixes or overlooked issues may later lead to zero-day exploits. In this paper, we propose $HYDRA$, a $Hy$brid heuristic-guided $D$eep $R$epresentation $A$rchitecture for predicting latent zero-day vulnerabilities in patched functions that combines rule-based heuristics with deep representation learning to detect latent risky code patterns that may persist after patches. It integrates static vulnerability rules, GraphCodeBERT embeddings, and a Variational Autoencoder (VAE) to uncover anomalies often missed by symbolic or neural models alone. We evaluate HYDRA in an unsupervised setting on patched functions from three diverse real-world software projects: Chrome, Android, and ImageMagick. Our results show HYDRA predicts 13.7%, 20.6%, and 24% of functions from Chrome, Android, and ImageMagick respectively as containing latent risks, including both heuristic matches and cases without heuristic matches ($None$) that may lead to zero-day vulnerabilities. It outperforms baseline models that rely solely on regex-derived features or their combination with embeddings, uncovering truly risky code variants that largely align with known heuristic patterns. These results demonstrate HYDRA's capability to surface hidden, previously undetected risks, advancing software security validation and supporting proactive zero-day vulnerabilities discovery.

The rapid expansion of the Internet of Things (IoT) is reshaping communication and operational practices across industries, but it also broadens the attack surface and increases susceptibility to security breaches. Artificial Intelligence has become a valuable solution in securing IoT networks, with Large Language Models (LLMs) enabling automated attack behavior analysis and mitigation suggestion in Network Intrusion Detection Systems (NIDS). Despite advancements, the use of LLMs in such systems further expands the attack surface, putting entire networks at risk by introducing vulnerabilities such as prompt injection and data poisoning. In this work, we attack an LLM-based IoT attack analysis and mitigation framework to test its adversarial robustness. We construct an attack description dataset and use it in a targeted data poisoning attack that applies word-level, meaning-preserving perturbations to corrupt the Retrieval-Augmented Generation (RAG) knowledge base of the framework. We then compare pre-attack and post-attack mitigation responses from the target model, ChatGPT-5 Thinking, to measure the impact of the attack on model performance, using an established evaluation rubric designed for human experts and judge LLMs. Our results show that small perturbations degrade LLM performance by weakening the linkage between observed network traffic features and attack behavior, and by reducing the specificity and practicality of recommended mitigations for resource-constrained devices.

The rapid proliferation of Internet of Things (IoT) devices has transformed numerous industries by enabling seamless connectivity and data-driven automation. However, this expansion has also exposed IoT networks to increasingly sophisticated security threats, including adversarial attacks targeting artificial intelligence (AI) and machine learning (ML)-based intrusion detection systems (IDS) to deliberately evade detection, induce misclassification, and systematically undermine the reliability and integrity of security defenses. To address these challenges, we propose a novel adversarial detection model that enhances the robustness of IoT IDS against adversarial attacks through SHapley Additive exPlanations (SHAP)-based fingerprinting. Using SHAP's DeepExplainer, we extract attribution fingerprints from network traffic features, enabling the IDS to reliably distinguish between clean and adversarially perturbed inputs. By capturing subtle attribution patterns, the model becomes more resilient to evasion attempts and adversarial manipulations. We evaluated the model on a standard IoT benchmark dataset, where it significantly outperformed a state-of-the-art method in detecting adversarial attacks. In addition to enhanced robustness, this approach improves model transparency and interpretability, thereby increasing trust in the IDS through explainable AI.

A decade after its academic introduction, RowHammer (RH) remains a moving target that continues to challenge both the industry and academia. With its potential to serve as a critical attack vector, the ever-decreasing RH threshold now threatens DRAM process technology scaling, with a superlinearly increasing cost of RH protection solutions. Due to their generality and relatively lower performance costs, architectural RH solutions are the first line of defense against RH. However, the field is fragmented with varying views of the problem, terminologies, and even threat models. In this paper, we systematize architectural RH defenses from the last decade through the lens of streaming algorithms. We provide a taxonomy that encompasses 48 different works. We map multiple architectural RH defenses to the classical streaming algorithms, which extends to multiple proposals that did not identify this link. We also provide two practitioner guides. The first guide analyzes which algorithm best fits a given RHTH, location, process technology, storage type, and mitigative action. The second guide encourages future research to consult existing algorithms when architecting RH defenses. We illustrate this by demonstrating how Reservoir-Sampling can improve related RH defenses, and also introduce StickySampling that can provide mathematical security that related studies do not guarantee.

Artificial intelligence (AI) agents are increasingly capable of initiating financial transactions on behalf of users or other agents. This evolution introduces a fundamental challenge: verifying both the authenticity of an autonomous agent and the true intent behind its transactions in a decentralized, trustless environment. Traditional payment systems assume human authorization, but autonomous, agent-led payments remove that safeguard. This paper presents a blockchain-based framework that cryptographically authenticates and verifies the intent of every AI-initiated transaction. The proposed system leverages decentralized identity (DID) standards and verifiable credentials to establish agent identities, on-chain intent proofs to record user authorization, and zero-knowledge proofs (ZKPs) to preserve privacy while ensuring policy compliance. Additionally, secure execution environments (TEE-based attestations) guarantee the integrity of agent reasoning and execution. The hybrid on-chain/off-chain architecture provides an immutable audit trail linking user intent to payment outcome. Through qualitative analysis, the framework demonstrates strong resistance to impersonation, unauthorized transactions, and misalignment of intent. This work lays the foundation for secure, auditable, and intent-aware autonomous economic agents, enabling a future of verifiable trust and accountability in AI-driven financial ecosystems.

Introducing Reliablocks, an on-chain reliability index for non-finalized blocks in Optimistic Rollups. This was built during the EigenLayer Infinite Hackathon at the Infinite Hacker House at DevCon 2024. As part of this research, we delivered a working Layer AVS WASMI component, a working Eigen Layer AVS component, EigenLayer Solidity smart contracts that work with the AVS component, a UI dashboard illustrating the reliability score and a derived interest rate for further utilization.

Efficiency and communication cost remain critical bottlenecks for practical Privacy-Preserving Machine Learning (PPML). Most existing frameworks rely on fixed-point arithmetic for strong security, which introduces significant precision loss and requires expensive cross-domain conversions (e.g., Arithmetic-to-Boolean) for non-linear operations. To address this, we propose PraxiMLP, a highly efficient three-party MLP framework grounded in practical security. The core of our work is a pair of novel additive-to-multiplicative conversion protocols that operate entirely within the arithmetic domain, thus avoiding expensive cross-domain conversions. By natively supporting loating-point numbers, PraxiMLP precisely handles non-linear functions, dramatically improving both efficiency and precision. Experimental results confirm that, compared to mainstream PPML frameworks, PraxiMLP delivers an average 8 orders of magnitude precision improvement on basic protocols and a 5x average model training speedup in a WAN environment.

Backdoor attacks in machine learning have drawn significant attention for their potential to compromise models stealthily, yet most research has focused on homogeneous data such as images. In this work, we propose a novel backdoor attack on tabular data, which is particularly challenging due to the presence of both numerical and categorical features. Our key idea is a novel technique to convert categorical values into floating-point representations. This approach preserves enough information to maintain clean-model accuracy compared to traditional methods like one-hot or ordinal encoding. By doing this, we create a gradient-based universal perturbation that applies to all features, including categorical ones. We evaluate our method on five datasets and four popular models. Our results show up to a 100% attack success rate in both white-box and black-box settings (including real-world applications like Vertex AI), revealing a severe vulnerability for tabular data. Our method is shown to surpass the previous works like Tabdoor in terms of performance, while remaining stealthy against state-of-the-art defense mechanisms. We evaluate our attack against Spectral Signatures, Neural Cleanse, Beatrix, and Fine-Pruning, all of which fail to defend successfully against it. We also verify that our attack successfully bypasses popular outlier detection mechanisms.

Federated Learning (FL) with Differential Privacy (DP) can ensure the protection of healthcare data privacy against malicious user attacks. However, FL with the DP method faces an inherent trade-off: stronger privacy protection degrades predictive accuracy due to injected noise. On the other hand, Homomorphic Encryption (HE) enables computations on encrypted data and the aggregation of encrypted gradients without introducing DP-induced noise, thereby potentially increasing the predictive accuracy. However, it may increase the computation cost. To improve the predictive accuracy while considering the computational ability of heterogeneous clients, we propose a Privacy-Preserving Machine Learning (PPML)-Hybrid method. In the proposed PPML-Hybrid method, clients distributedly select either HE or DP based on their computational resources, so that HE clients contribute noise-free updates while DP clients reduce computational overhead. Meanwhile, clients with high computational resources can flexibly adopt HE or DP according to their privacy needs. Performance evaluation on omics datasets shows that our proposed method achieves comparable predictive accuracy while significantly reducing computation time relative to HE-only. Additionally, it outperforms DP-only methods under equivalent or stricter privacy budgets.