Papers

In this paper, we present the first large-scale study exploring whether JavaScript code generated by Large Language Models (LLMs) can reveal which model produced it, enabling reliable authorship attribution and model fingerprinting. With the rapid rise of AI-generated code, attribution is playing a critical role in detecting vulnerabilities, flagging malicious content, and ensuring accountability. While AI-vs-human detection usually treats AI as a single category we show that individual LLMs leave unique stylistic signatures, even among models belonging to the same family or parameter size. To this end, we introduce LLM-NodeJS, a dataset of 50,000 Node.js back-end programs from 20 large language models. Each has four transformed variants, yielding 250,000 unique JavaScript samples and two additional representations (JSIR and AST) for diverse research applications. Using this dataset, we benchmark traditional machine learning classifiers against fine-tuned Transformer encoders and introduce CodeT5-JSA, a custom architecture derived from the 770M-parameter CodeT5 model with its decoder removed and a modified classification head. It achieves 95.8% accuracy on five-class attribution, 94.6% on ten-class, and 88.5% on twenty-class tasks, surpassing other tested models such as BERT, CodeBERT, and Longformer. We demonstrate that classifiers capture deeper stylistic regularities in program dataflow and structure, rather than relying on surface-level features. As a result, attribution remains effective even after mangling, comment removal, and heavy code transformations. To support open science and reproducibility, we release the LLM-NodeJS dataset, Google Colab training scripts, and all related materials on GitHub: https://github.com/LLM-NodeJS-dataset.

Open-source large language models (LLMs) have demonstrated considerable dominance over proprietary LLMs in resolving neural processing tasks, thanks to the collaborative and sharing nature. Although full access to source codes, model parameters, and training data lays the groundwork for transparency, we argue that such a full-access manner is vulnerable to MEAs, and their ill-effects are not fully understood. In this paper, we conduct a systematic formalization for MEAs on open-source LLMs by enumerating all possible threat models associated with adversary objectives, knowledge, and capabilities. Therein, the threat posed by adversaries with internal knowledge, who inject payloads and triggers during the model sharing phase, is of practical interest. We go even further and propose the first MEA against open-source LLMs, dubbed MEASER, which wields impacts through identifying targeted parameters, embedding payloads, injecting triggers, and executing payloads sequentially. Particularly, MEASER enhances the attack robustness against quantization and parameter-efficient fine-tuning (PEFT) by employing the Magnitude-Adaptive Relative Quantization Index Modulation (MAR-QIM) mechanism, synergized with LDPC codes and spread spectrum modulation. In addition, to achieve stealthiness, MEASER devises the performance-aware importance metric to identify targeted parameters with the least degradation of model performance. Extensive experiments on four popular open-source LLMs show that the stealth rate of MEASER outperforms existing MEAs (for general DNNs) significantly, while consistently achieving a 0 bit error rate (BER) in all settings. Moreover, MEASER also maintains superior stealthiness on quantized models. We appeal for investigations on countermeasures against MEASER in view of the significant attack effectiveness.

Post-quantum cryptography (PQC) is moving from evaluation to deployment as NIST finalizes standards for ML-KEM, ML-DSA, and SLH-DSA. This survey maps the space from foundations to practice. We first develop a taxonomy across lattice-, code-, hash-, multivariate-, isogeny-, and MPC-in-the-Head families, summarizing security assumptions, cryptanalysis, and standardization status. We then compare performance and communication costs using representative, implementation-grounded measurements, and review hardware acceleration (AVX2, FPGA/ASIC) and implementation security with a focus on side-channel resistance. Building upward, we examine protocol integration (TLS, DNSSEC), PKI and certificate hygiene, and deployment in constrained and high-assurance environments (IoT, cloud, finance, blockchain). We also discuss complementarity with quantum technologies (QKD, QRNGs) and the limits of near-term quantum computing. Throughout, we emphasize crypto-agility, hybrid migration, and evidence-based guidance for operators. We conclude with open problems spanning parameter agility, leakage-resilient implementations, and domain-specific rollout playbooks. This survey aims to be a practical reference for researchers and practitioners planning quantum-safe systems, bridging standards, engineering, and operations.

GraphQL's flexible query model and nested data dependencies expose APIs to complex, context-dependent vulnerabilities that are difficult to uncover using conventional testing tools. Existing fuzzers either rely on random payload generation or rigid mutation heuristics, failing to adapt to the dynamic structures of GraphQL schemas and responses. We present PrediQL, the first retrieval-augmented, LLM-guided fuzzer for GraphQL APIs. PrediQL combines large language model reasoning with adaptive feedback loops to generate semantically valid and diverse queries. It models the choice of fuzzing strategy as a multi-armed bandit problem, balancing exploration of new query structures with exploitation of past successes. To enhance efficiency, PrediQL retrieves and reuses execution traces, schema fragments, and prior errors, enabling self-correction and progressive learning across test iterations. Beyond input generation, PrediQL integrates a context-aware vulnerability detector that uses LLM reasoning to analyze responses, interpreting data values, error messages, and status codes to identify issues such as injection flaws, access-control bypasses, and information disclosure. Our evaluation across open-source and benchmark GraphQL APIs shows that PrediQL achieves significantly higher coverage and vulnerability discovery rates compared to state-of-the-art baselines. These results demonstrate that combining retrieval-augmented reasoning with adaptive fuzzing can transform API security testing from reactive enumeration to intelligent exploration.

OpenAI (2025) showed that training against a chain of thought (CoT) monitor can cause obfuscated CoTs, which contain bad behavior the monitor cannot detect. They proposed to keep CoTs monitorable by training only against output monitors that do not have access to CoT. We show that such training can still cause obfuscated CoTs via two mechanisms. First, when a model is trained to produce a safe-looking output, that model may generalize to making its CoTs look safe. Second, since later tokens are conditioned on earlier ones, safe-looking CoTs may increase the likelihood of safe outputs, causing safe-looking CoTs to be reinforced. We introduce two mitigations to address these two issues, which achieve a Pareto improvement in terms of monitorability and task performance compared to regular training.

Knowledge poisoning poses a critical threat to Retrieval-Augmented Generation (RAG) systems by injecting adversarial content into knowledge bases, tricking Large Language Models (LLMs) into producing attacker-controlled outputs grounded in manipulated context. Prior work highlights LLMs' susceptibility to misleading or malicious retrieved content. However, real-world fact-checking scenarios are more challenging, as credible evidence typically dominates the retrieval pool. To investigate this problem, we extend knowledge poisoning to the fact-checking setting, where retrieved context includes authentic supporting or refuting evidence. We propose ADMIT (ADversarial Multi-Injection Technique), a few-shot, semantically aligned poisoning attack that flips fact-checking decisions and induces deceptive justifications, all without access to the target LLMs, retrievers, or token-level control. Extensive experiments show that ADMIT transfers effectively across 4 retrievers, 11 LLMs, and 4 cross-domain benchmarks, achieving an average attack success rate (ASR) of 86% at an extremely low poisoning rate of $0.93 \times 10^{-6}$, and remaining robust even in the presence of strong counter-evidence. Compared with prior state-of-the-art attacks, ADMIT improves ASR by 11.2% across all settings, exposing significant vulnerabilities in real-world RAG-based fact-checking systems.

The rapid spread of text generated by large language models (LLMs) makes it increasingly difficult to distinguish authentic human writing from machine output. Watermarking offers a promising solution: model owners can embed an imperceptible signal into generated text, marking its origin. Most leading approaches seed an LLM's next-token sampling with a pseudo-random key that can later be recovered to identify the text as machine-generated, while only minimally altering the model's output distribution. However, these methods suffer from two related issues: (i) watermarks are brittle to simple surface-level edits such as paraphrasing or reordering; and (ii) adversaries can append unrelated, potentially harmful text that inherits the watermark, risking reputational damage to model owners. To address these issues, we introduce SimKey, a semantic key module that strengthens watermark robustness by tying key generation to the meaning of prior context. SimKey uses locality-sensitive hashing over semantic embeddings to ensure that paraphrased text yields the same watermark key, while unrelated or semantically shifted text produces a different one. Integrated with state-of-the-art watermarking schemes, SimKey improves watermark robustness to paraphrasing and translation while preventing harmful content from false attribution, establishing semantic-aware keying as a practical and extensible watermarking direction.

Software vulnerabilities remain a persistent risk, yet static and dynamic analyses often overlook structural dependencies that shape insecure behaviors. Viewing programs as heterogeneous graphs, we capture control- and data-flow relations as complex interaction networks. Our hybrid framework combines these graph representations with light-weight (<4B) local LLMs, uniting topological features with semantic reasoning while avoiding the cost and privacy concerns of large cloud models. Evaluated on Java vulnerability detection (binary classification), our method achieves 93.57% accuracy-an 8.36% gain over Graph Attention Network-based embeddings and 17.81% over pretrained LLM baselines such as Qwen2.5 Coder 3B. Beyond accuracy, the approach extracts salient subgraphs and generates natural language explanations, improving interpretability for developers. These results pave the way for scalable, explainable, and locally deployable tools that can shift vulnerability analysis from purely syntactic checks to deeper structural and semantic insights, facilitating broader adoption in real-world secure software development.

Since being proposed in 2006, differential privacy has become a standard method for quantifying certain risks in publishing or sharing analyses of sensitive data. At its heart, differential privacy measures risk in terms of the differences between probability distributions, which is a central topic in information theory. A differentially private algorithm is a channel between the underlying data and the output of the analysis. Seen in this way, the guarantees made by differential privacy can be understood in terms of properties of this channel. In this article we examine a few of the key connections between information theory and the formulation/application of differential privacy, giving an “operational significance” for relevant information measures.

The integration of Large Language Models (LLMs) into computer applications has introduced transformative capabilities but also significant security challenges. Existing safety alignments, which primarily focus on semantic interpretation, leave LLMs vulnerable to attacks that use non-standard data representations. This paper introduces ArtPerception, a novel black-box jailbreak framework that strategically leverages ASCII art to bypass the security measures of state-of-the-art (SOTA) LLMs. Unlike prior methods that rely on iterative, brute-force attacks, ArtPerception introduces a systematic, two-phase methodology. Phase 1 conducts a one-time, model-specific pre-test to empirically determine the optimal parameters for ASCII art recognition. Phase 2 leverages these insights to launch a highly efficient, one-shot malicious jailbreak attack. We propose a Modified Levenshtein Distance (MLD) metric for a more nuanced evaluation of an LLM's recognition capability. Through comprehensive experiments on four SOTA open-source LLMs, we demonstrate superior jailbreak performance. We further validate our framework's real-world relevance by showing its successful transferability to leading commercial models, including GPT-4o, Claude Sonnet 3.7, and DeepSeek-V3, and by conducting a rigorous effectiveness analysis against potential defenses such as LLaMA Guard and Azure's content filters. Our findings underscore that true LLM security requires defending against a multi-modal space of interpretations, even within text-only inputs, and highlight the effectiveness of strategic, reconnaissance-based attacks. Content Warning: This paper includes potentially harmful and offensive model outputs.

Unlike regular tokens derived from existing text corpora, special tokens are artificially created to annotate structured conversations during the fine-tuning process of Large Language Models (LLMs). Serving as metadata of training data, these tokens play a crucial role in instructing LLMs to generate coherent and context-aware responses. We demonstrate that special tokens can be exploited to construct four attack primitives, with which malicious users can reliably bypass the internal safety alignment of online LLM services and circumvent state-of-the-art (SOTA) external content moderation systems simultaneously. Moreover, we found that addressing this threat is challenging, as aggressive defense mechanisms-such as input sanitization by removing special tokens entirely, as suggested in academia-are less effective than anticipated. This is because such defense can be evaded when the special tokens are replaced by regular ones with high semantic similarity within the tokenizer's embedding space. We systemically evaluated our method, named MetaBreak, on both lab environment and commercial LLM platforms. Our approach achieves jailbreak rates comparable to SOTA prompt-engineering-based solutions when no content moderation is deployed. However, when there is content moderation, MetaBreak outperforms SOTA solutions PAP and GPTFuzzer by 11.6% and 34.8%, respectively. Finally, since MetaBreak employs a fundamentally different strategy from prompt engineering, the two approaches can work synergistically. Notably, empowering MetaBreak on PAP and GPTFuzzer boosts jailbreak rates by 24.3% and 20.2%, respectively.

System passwords serve as critical credentials for user authentication and access control when logging into operating systems or applications. Upon entering a valid password, users pass verification to access system resources and execute corresponding operations. In recent years, frequent password cracking attacks targeting system passwords have posed a severe threat to information system security. To address this challenge, in-depth research into password cracking attack methods and defensive technologies holds significant importance. This paper conducts systematic research on system password security, focusing on analyzing typical password cracking methods such as brute force attacks, dictionary attacks, and rainbow table attacks, while evaluating the effectiveness of existing defensive measures. The experimental section utilizes common cryptanalysis tools, such as John the Ripper and Hashcat, to simulate brute force and dictionary attacks. Five test datasets, each generated using Message Digest Algorithm 5 (MD5), Secure Hash Algorithm 256-bit (SHA 256), and bcrypt hash functions, are analyzed. By comparing the overall performance of different hash algorithms and password complexity strategies against these attacks, the effectiveness of defensive measures such as salting and slow hashing algorithms is validated. Building upon this foundation, this paper further evaluates widely adopted defense mechanisms, including account lockout policies, multi-factor authentication, and risk adaptive authentication. By integrating experimental data with recent research findings, it analyzes the strengths and limitations of each approach while proposing feasible improvement recommendations and optimization strategies.

Deep learning-based weather forecasting (DLWF) models leverage past weather observations to generate future forecasts, supporting a wide range of downstream applications, including tropical cyclone (TC) prediction. In this paper, we investigate their vulnerability to adversarial attacks, where subtle perturbations to the upstream forecasts can alter the downstream TC trajectory predictions. Although research into adversarial attacks on DLWF models has grown recently, it remains challenging to craft perturbed upstream forecasts that steer the downstream outputs toward attacker-specified trajectories. First, conventional TC detection systems are opaque, non-differentiable black boxes, making standard gradient-based attacks infeasible. Second, the extreme rarity of TC events leads to severe class imbalance problem, making it difficult to develop attack methods for perturbing upstream forecasts that produce realistic-looking cyclone paths aligned with attacker's target trajectories. To overcome these limitations, we propose Cyc-Attack, a novel method for perturbing the upstream forecasts of DLWF models to generate adversarial trajectories. The proposed method uses a differentiable surrogate model to approximate the TC detector's output, enabling the application of gradient-based attacks. Cyc-Attack also employs a skewness-aware loss function with kernel dilation strategy to address the imbalance problem. Finally, a distance-based gradient weighting scheme and regularization are used to constrain the perturbations and eliminate unrealistic-looking trajectories, thereby making the adversarial upstream forecasts less easily detectable. Our experiments show that Cyc-Attack achieves a higher true positive rate in matching the attacker's target trajectories, along with lower false alarm rates and stealthier perturbations than conventional attack methods.

Advances in image tampering pose serious security threats, underscoring the need for effective image manipulation localization (IML). While supervised IML achieves strong performance, it depends on costly pixel-level annotations. Existing weakly supervised or training-free alternatives often underperform and lack interpretability. We propose the In-Context Forensic Chain (ICFC), a training-free framework that leverages multi-modal large language models (MLLMs) for interpretable IML tasks. ICFC integrates an objectified rule construction with adaptive filtering to build a reliable knowledge base and a multi-step progressive reasoning pipeline that mirrors expert forensic workflows from coarse proposals to fine-grained forensics results. This design enables systematic exploitation of MLLM reasoning for image-level classification, pixel-level localization, and text-level interpretability. Across multiple benchmarks, ICFC not only surpasses state-of-the-art training-free methods but also achieves competitive or superior performance compared to weakly and fully supervised approaches.

Harmful fine-tuning issues present significant safety challenges for fine-tuning-as-a-service in large language models. Existing alignment-stage defenses, e.g., Vaccine, Repnoise, Booster, and T-Vaccine, mitigate harmful fine-tuning issues by enhancing the model's robustness during the alignment phase. While these methods have been proposed to mitigate the issue, they often overlook a critical upstream factor: the role of the original safety-alignment data. We observe that their defense performance and computational efficiency remain constrained by the quality and composition of the alignment dataset. To address this limitation, we propose Pharmacist, a safety alignment data curation solution that enhances defense against harmful fine-tuning by selecting a high-quality and safety-critical core subset from the original alignment data. The core idea of Pharmacist is to train an alignment data selector to rank alignment data. Specifically, up-ranking high-quality and safety-critical alignment data, down-ranking low-quality and non-safety-critical data. Empirical results indicate that models trained on datasets selected by Pharmacist outperform those trained on datasets selected by existing selection methods in both defense and inference performance. In addition, Pharmacist can be effectively integrated with mainstream alignment-stage defense methods. For example, when applied to RepNoise and T-Vaccine, using the dataset selected by Pharmacist instead of the full dataset leads to improvements in defense performance by 2.60% and 3.30%, respectively, and enhances inference performance by 3.50% and 1.10%. Notably, it reduces training time by 56.83% and 57.63%, respectively. Our code is available at https://github.com/Lslland/Pharmacist.

Large vision-language model (LVLM)-based web agents are emerging as powerful tools for automating complex online tasks. However, when deployed in real-world environments, they face serious security risks, motivating the design of security evaluation benchmarks. Existing benchmarks provide only partial coverage, typically restricted to narrow scenarios such as user-level prompt manipulation, and thus fail to capture the broad range of agent vulnerabilities. To address this gap, we present \tool{}, the first holistic benchmark for evaluating the security of LVLM-based web agents. \tool{} first introduces a unified evaluation suite comprising six simulated but realistic web environments (\eg, e-commerce platforms, community forums) and includes 2,970 high-quality trajectories spanning diverse tasks and attack settings. The suite defines a structured taxonomy of six attack vectors spanning both user-level and environment-level manipulations. In addition, we introduce a multi-layered evaluation protocol that analyzes agent failures across three critical dimensions: internal reasoning, behavioral trajectory, and task outcome, facilitating a fine-grained risk analysis that goes far beyond simple success metrics. Using this benchmark, we conduct large-scale experiments on 9 representative LVLMs, which fall into three categories: general-purpose, agent-specialized, and GUI-grounded. Our results show that all tested agents are consistently vulnerable to subtle adversarial manipulations and reveal critical trade-offs between model specialization and security. By providing (1) a comprehensive benchmark suite with diverse environments and a multi-layered evaluation pipeline, and (2) empirical insights into the security challenges of modern LVLM-based web agents, \tool{} establishes a foundation for advancing trustworthy web agent deployment.

Machine learning has become a crucial part of our lives, with applications spanning nearly every aspect of our daily activities. However, using personal information in machine learning applications has sparked significant security and privacy concerns about user data. To address these challenges, different privacy-preserving machine learning (PPML) frameworks have been developed to protect sensitive information in machine learning applications. These frameworks generally attempt to balance design trade-offs such as computational efficiency, communication overhead, security guarantees, and scalability. Despite the advancements, selecting the optimal framework and parameters for specific deployment scenarios remains a complex and critical challenge for privacy and security application developers. We present Prismo, an open-source recommendation system designed to aid in selecting optimal parameters and frameworks for different PPML application scenarios. Prismo enables users to explore a comprehensive space of PPML frameworks through various properties based on user-defined objectives. It supports automated filtering of suitable candidate frameworks by considering parameters such as the number of parties in multi-party computation or federated learning and computation cost constraints in homomorphic encryption. Prismo models every use case into a Linear Integer Programming optimization problem, ensuring tailored solutions are recommended for each scenario. We evaluate Prismo's effectiveness through multiple use cases, demonstrating its ability to deliver best-fit solutions in different deployment scenarios.

Software systems can be represented as graphs, capturing dependencies among functions and processes. An interesting aspect of software systems is that they can be represented as different types of graphs, depending on the extraction goals and priorities. For example, function calls within the software can be captured to create function call graphs, which highlight the relationships between functions and their dependencies. Alternatively, the processes spawned by the software can be modeled to generate process interaction graphs, which focus on runtime behavior and inter-process communication. While these graph representations are related, each captures a distinct perspective of the system, providing complementary insights into its structure and operation. While previous studies have leveraged graph neural networks (GNNs) to analyze software behaviors, most of this work has focused on a single type of graph representation. The joint modeling of both function call graphs and process interaction graphs remains largely underexplored, leaving opportunities for deeper, multi-perspective analysis of software systems. This paper presents a pipeline for constructing and training Function Call Graphs (FCGs) and Process Call Graphs (PCGs) and learning joint embeddings. We demonstrate that joint embeddings outperform a single-graph model. In this paper, we propose GeminiNet, a unified neural network approach that learns joint embeddings from both FCGs and PCGs. We construct a new dataset of 635 Windows executables (318 malicious and 317 benign), extracting FCGs via Ghidra and PCGs via Any.Run sandbox. GeminiNet employs dual graph convolutional branches with an adaptive gating mechanism that balances contributions from static and dynamic views.

Contemporary web application architectures involve many layers of proxy services that process traffic. Due to the complexity of HTTP and vendor design decisions, these proxies sometimes process a given request in different ways. Attackers can exploit these processing discrepancies to launch damaging attacks including web cache poisoning and request smuggling. Discrepancy attacks are surging, yet, there exists no systemic defense. In this work, we propose the first comprehensive defense to address this problem, called HTTP Request Synchronization. Our scheme uses standard HTTP extension mechanisms to augment each request with a complete processing history. It propagates this context through the traffic path detailing how each server hop has processed said request. Using this history, every proxy server can validate that their processing is consistent with all previous hops, eliminating discrepancy attacks. We implement our scheme for 5 popular proxy technologies, Apache, NGINX, HAProxy, Varnish, and Cloudflare, demonstrating its practical impact.

Deep learning-based radio frequency fingerprinting (RFFP) has become an enabling physical-layer security technology, allowing device identification and authentication through received RF signals. This technology, however, faces significant challenges when it comes to adapting to domain variations, such as time, location, environment, receiver and channel. For Bluetooth Low Energy (BLE) devices, addressing these challenges is particularly crucial due to the BLE protocol's frequency-hopping nature. In this work, and for the first time, we investigated the frequency hopping effect on RFFP of BLE devices, and proposed a novel, low-cost, domain-adaptive feature extraction method. Our approach improves the classification accuracy by up to 58% across environments and up to 80% across receivers compared to existing benchmarks.

Multi-modal threat detection faces a fundamental challenge that involves security tools operating in isolation, and this creates streams of network, email, and system data with no natural alignment or correlation. We present Hierarchical Multi-Modal Threat Intelligence Fusion (HM-TIF), a framework explicitly designed for this realistic scenario where naturally aligned multi-modal attack data does not exist. Unlike prior work that assumes or creates artificial alignment, we develop principled methods for correlating independent security data streams while maintaining operational validity. Our architecture employs hierarchical cross-attention with dynamic weighting that adapts to data availability and threat context, coupled with a novel temporal correlation protocol that preserves statistical independence. Evaluation on UNSW-NB15, CSE-CIC-IDS2018, and CICBell-DNS2021 datasets demonstrates that HM-TIF achieves 88.7% accuracy with a critical 32% reduction in false positive rates, even without true multi-modal training data. The framework maintains robustness when modalities are missing, making it immediately deployable in real security operations where data streams frequently have gaps. Our contributions include: (i) the first multi-modal security framework explicitly designed for non-aligned data, (ii) a temporal correlation protocol that avoids common data leakage pitfalls, (iii) empirical validation that multi-modal fusion provides operational benefits even without perfect alignment, and (iv) practical deployment guidelines for security teams facing heterogeneous, uncoordinated data sources. Index Terms: multi-modal learning, threat intelligence, non-aligned data, operational security, cross-attention mechanisms, practical deployment

Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal perception and generation, yet their safety alignment remains a critical challenge.Existing defenses and vulnerable to multimodal jailbreaks, as visual inputs introduce new attack surfaces, reasoning chains lack safety supervision, and alignment often degrades under modality fusion.To overcome these limitation, we propose VisuoAlign, a framework for multi-modal safety alignment via prompt-guided tree search.VisuoAlign embeds safety constrains into the reasoning process through visual-textual interactive prompts, employs Monte Carlo Tree Search(MCTS) to systematically construct diverse safety-critical prompt trajectories, and introduces prompt-based scaling to ensure real-time risk detection and compliant responses.Extensive experiments demonstrate that VisuoAlign proactively exposes risks, enables comprehensive dataset generation, and significantly improves the robustness of LVLMs against complex cross-modal threats.

Internet memes have emerged as a popular multimodal medium, yet they are increasingly weaponized to convey harmful opinions through subtle rhetorical devices like irony and metaphor. Existing detection approaches, including Multimodal Large Language Model (MLLM)-based techniques, struggle with these implicit expressions, leading to frequent misjudgments. This paper introduces PatMD, a novel approach that detects harmful memes by learning from and proactively mitigating these potential misjudgment risks. Our core idea is to move beyond superficial content-level matching and instead identify the underlying misjudgment risk patterns, proactively guiding the MLLMs to avoid known misjudgment pitfalls. We first construct a knowledge base where each meme is deconstructed into a misjudgment risk pattern explaining why it might be misjudged, either overlooking harmful undertones (false negative) or overinterpreting benign content (false positive). For a given target meme, PatMD retrieves relevant patterns and utilizes them to dynamically guide the MLLM's reasoning. Experiments on a benchmark of 6,626 memes across 5 harmful detection tasks show that PatMD outperforms state-of-the-art baselines, achieving an average of 8.30% improvement in F1-score and 7.71% improvement in accuracy, while exhibiting consistent robustness on unseen and adversarial memes.

This paper addresses one of the most noteworthy issues in the recent virtual asset market, the privacy concerns related to token transactions of Real-World Assets tokens, known as RWA tokens. Following the advent of Bitcoin, the virtual asset market has experienced explosive growth, spawning movements to link real-world assets with virtual assets. However, due to the transparency principle of blockchain technology, the anonymity of traders cannot be guaranteed. In the existing blockchain environment, there have been instances of protecting the privacy of fungible tokens (FTs) using mixer services. Moreover, numerous studies have been conducted to secure the privacy of non-fungible tokens (NFTs). However, due to the unique characteristics of RWA tokens and the limitations of each study, it has been challenging to achieve the goal of anonymity protection effectively. This paper proposes a new token trading platform, the ARTeX, designed to resolve these issues. This platform not only addresses the shortcomings of existing methods but also ensures the anonymity of traders while enhancing safeguards against illegal activities.

We argue that the principal application for blockchain technology will not be in the financial sector, but rather in maintaining decentralized human governance, from archives to transparent policies encoded in the blockchain in the form of smart contracts.. Such decentralized, blockchain-grounded governance comes not a moment too soon, as nation states are dissolving before our eyes. Will blockchain-based communities replace the nation state? What are the prospects and dangers of this development?

This paper investigates the use of synthetic face data to enhance Single-Morphing Attack Detection (S-MAD), addressing the limitations of availability of large-scale datasets of bona fide images due to privacy concerns. Various morphing tools and cross-dataset evaluation schemes were utilized to conduct this study. An incremental testing protocol was implemented to assess the generalization capabilities as more and more synthetic images were added. The results of the experiments show that generalization can be improved by carefully incorporating a controlled number of synthetic images into existing datasets or by gradually adding bona fide images during training. However, indiscriminate use of synthetic data can lead to sub-optimal performance. Evenmore, the use of only synthetic data (morphed and non-morphed images) achieves the highest Equal Error Rate (EER), which means in operational scenarios the best option is not relying only on synthetic data for S-MAD.

Fingerprinting radio frequency (RF) emitters typically involves finding unique characteristics that are featured in their received signal. These fingerprints are nuanced, but sufficiently detailed, motivating the pursuit of methods that can successfully extract them. The downstream task that requires the most meticulous RF fingerprinting (RFF) is known as specific emitter identification (SEI), which entails recognising each individual transmitter. RFF and SEI have a long history, with numerous defence and civilian applications such as signal intelligence, electronic surveillance, physical-layer authentication of wireless devices, to name a few. In recent years, data-driven RFF approaches have become popular due to their ability to automatically learn intricate fingerprints. They generally deliver superior performance when compared to traditional RFF techniques that are often labour-intensive, inflexible, and only applicable to a particular emitter type or transmission scheme. In this paper, we present a generic and versatile machine learning (ML) framework for data-driven RFF with several popular downstream tasks such as SEI, data association (EDA) and RF emitter clustering (RFEC). It is emitter-type agnostic. We then demonstrate the introduced framework for several tasks using real RF datasets for spaceborne surveillance, signal intelligence and countering drones applications.

Secret-key generation and agreement based on wireless channel reciprocity offers a promising avenue for securing IoT networks. However, existing approaches predominantly rely on the similarity of instantaneous channel measurement samples between communicating devices. This narrow view of reciprocity is often impractical, as it is highly susceptible to noise, asynchronous sampling, channel fading, and other system-level imperfections – all of which significantly impair key generation performance. Furthermore, the quantization step common in traditional schemes introduces irreversible errors, further limiting efficiency. In this work, we propose a novel approach for secret-key generation by using wavelet scattering networks to extract robust and reciprocal CSI features. Dimensionality reduction is applied to uncover hidden cluster structures, which are then used to build hidden Markov models for efficient key agreement. Our approach eliminates the need for quantization and effectively captures channel randomness. It achieves a 5x improvement in key generation rate compared to traditional benchmarks, providing a secure and efficient solution for key generation in resource-constrained IoT environments.

Proof of Work (PoW) is widely regarded as the most secure permissionless blockchain consensus protocol. However, its reliance on computationally intensive yet externally useless puzzles results in excessive electric energy wasting. To alleviate this, Proof of Useful Work (PoUW) has been explored as an alternative to secure blockchain platforms while also producing real-world value. Despite this promise, existing PoUW proposals often fail to embed the integrity of the chain and the identity of the miner into the puzzle solutions, not meeting the necessary requirements for PoW and thus rendering them vulnerable. In this work, we propose a PoUW consensus protocol that computes client-outsourced SNARK proofs as a byproduct, which are simultaneously used to secure the consensus protocol. We further leverage this mechanism to design a decentralized marketplace for outsourcing SNARK proof generation, which is, to the best of our knowledge, the first such marketplace operating at the consensus layer while meeting all necessary properties of PoW.

With the proliferation of decentralized applications (DApps), the conflict between the transparency of blockchain technology and user data privacy has become increasingly prominent. While Decentralized Identity (DID) and Verifiable Credentials (VCs) provide a standardized framework for user data sovereignty, achieving trusted identity verification and data sharing without compromising privacy remains a significant challenge. This paper proposes a novel, comprehensive framework that integrates DIDs and VCs with efficient Zero-Knowledge Proof (ZKP) schemes to address this core issue. The key contributions of this framework are threefold: first, it constructs a set of strong privacy-preserving protocols based on zk-STARKs, allowing users to prove that their credentials satisfy specific conditions (e.g., "age is over 18") without revealing any underlying sensitive data. Second, it designs a scalable, privacy-preserving credential revocation mechanism based on cryptographic accumulators, effectively solving credential management challenges in large-scale scenarios. Finally, it integrates a practical social key recovery scheme, significantly enhancing system usability and security. Through a prototype implementation and performance evaluation, this paper quantitatively analyzes the framework's performance in terms of proof generation time, verification overhead, and on-chain costs. Compared to existing state-of-the-art systems based on zk-SNARKs, our framework, at the cost of a larger proof size, significantly improves prover efficiency for complex computations and provides stronger security guarantees, including no trusted setup and post-quantum security. Finally, a case study in the decentralized finance (DeFi) credit scoring scenario demonstrates the framework's immense potential for unlocking capital efficiency and fostering a trusted data economy.