Zeinab Alipanahloo, Michael Duchesne, Kaiwen Zhang
RzkFL is an end-to-end, privacy-preserving machine-learning framework that fuses Federated Learning (FL) with recursive zero-knowledge proofs (ZKPs) to protect data, models, and users while unlocking verifiable inference. Models are trained entirely on local devices, so sensitive data never leave the premises. The resulting model can be monetized by offering verifiable predictions on a pay-per-use basis. During inference, each customer independently computes predictions using private data, making it essential to verify that these inference results are computed correctly and honestly. Unlike existing approaches that rely on heavy communication or centralized trust assumptions, RzkFL allows each customer to generate a cryptographic proof of correct local inference, which can be succinctly verified without revealing input data or model parameters either by the customer or a third party. The core innovation lies in the use of recursive ZKPs, enabling each customer to generate small, composable proofs for intermediate layers of neural network inference. These proofs are then recursively aggregated into a single succinct proof using the Nova proof folding scheme. Nova’s design eliminates the traditional sequential dependency of recursive proofs by enabling incrementally verifiable computation through a folding scheme. RzkFL supports on-chain verification via Ethereum smart contracts, allowing AI results to flow directly into financial workflows. A decentralized file storage system maintains the integrity and availability of the global model. We introduce specialized circuits for input, hidden, and output layers to optimize proof generation time and gas costs. The customer can generate proof for the entire inference computation or delegate the proof generation for the intermediate layers and the output layer to another party. The design suits privacy-preserving machine learning scenarios where customer devices are resource-constrained. Our results show that RzkFL can significantly reduce proof size and verification costs while maintaining privacy, integrity, and scalability in federated inference. This makes it a compelling approach for real-world decentralized AI systems requiring strong verifiability guarantees.
Tuan-Dung Tran, Dinh Khang Nguyen, Quang Trung Do, Van-Hau Pham
Cross-chain bridges, while critical for interoperability in the Web3 ecosystem, have become a primary target for exploits, accounting for over $4.3 billion in losses—nearly 40% of all value stolen in recent years. Existing security paradigms fail to address this threat adequately due to a fundamental trade-off between pre-deployment static analysis and real-time dynamic monitoring. Current security paradigms are trapped in a critical trade-off: pre-deployment static analysis lacks runtime context and suffers from high false-positives, while real-time dynamic monitoring is blind to the underlying source-code vulnerabilities that enable sophisticated attacks. This paper introduces VeriBridge, a novel framework that breaks this impasse. VeriBridge pioneers a synergistic fusion of static intelligence and dynamic graph learning. It enriches real-time transaction graphs with fine-grained vulnerability data extracted from static analysis, providing crucial security context to an unsupervised Graph Autoencoder. By learning a high-fidelity model of normal behavior, VeriBridge detects malicious transactions, including zero-day exploits, as significant deviations from this learned norm, identified by high reconstruction error. Evaluated on a comprehensive dataset of real-world attacks, including the Poly Network and THORchain exploits, VeriBridge achieves a 96% F1-score, demonstrating a new frontier in robust, real-time security for critical blockchain infrastructure.
Decentralized finance (DeFi) protocols are becoming increasingly targeted by cyber threats, such as liquidity drain attacks, smart contracts flaws that leverage instant loans, and increasingly sophisticated threats that include DarkGate ransomware. We develop a hybrid framework that integrates CTI and predictive analytics to facilitate improving consensus mechanisms in a blockchain network. The proposed framework is centered on three layers , a data collection and processing layer, a security oracle layer that engages to mitigate intervention, and a dynamic adaptive mechanism to reach consensus. A 250-node testbed was built and deployed with the Hyperledger Besu and Geth deployments of Ethereum incorporating hybrid GRU-BiLSTM which utilize GNN's for predicting attacks. The results reveal improvements of transaction processing TPS of up to +236%, settlement latency improved -75%, fork rate improved to less than 3%, and downtime improved from 15% to 1.5%. Statistical tests T-Test and ANOVA also reveal these were of high statistically significance at p < 0.01. This study emphasizes that bridging functional aspects of AI with adaptive consensus mechanisms will be an effective approach at combating advanced cyber-attacks while maintaining reliability and resilience in DeFi systems.
Smart contracts play a pivotal role in decentralized applications but are subject to security vulnerabilities often difficult to detect. Traditional static and symbolic analysis tools cannot handle intricate logic and are limited in adaptability and explainability. Recent development of large language models (LLMs) provide new opportunities for vulnerability detection, but single-model methods often suffer from inconsistency and prompt sensitivity. This paper introduces a collaborative LLM-based model that enhances detection robustness through semantic similarity-based few-shot prompting and multi-LLM reasoning. Our model integrates diverse LLMs (ChatGPT, Gemini, Grok) as worker nodes, along with an aggregator model to resolve disagreements via justification analysis and final prediction consolidation. Experimental evaluations on the SmartBugs benchmark demonstrated a remarkable enhancement in detection accuracy (96.25%) and response time compared to other models. The proposed model provides a scalable and explainable solution for smart contract auditing, illustrating the strength of LLM collaboration in security-critical applications.
The rapid expansion of multi-cloud ecosystems has intensified the demand for privacy-preserving analytics across untrusted infrastructures. This paper proposes Federated Zero-Trust Analytics (FZTA), a framework that integrates federated learning, zero-trust security, and privacy-enhancing computation to enable secure data collaboration without centralized trust. The design combines continuous identity verification, decentralized policy enforcement, and hybrid cryptography based on homomorphic encryption and differential privacy. Evaluation across three commercial clouds demonstrates that FZTA achieves near baseline model accuracy (within 2% of centralized training) while maintaining (ε&lt;1.2, δ=10−5) differential privacy guarantees and less than 20% computational overhead. The framework resists eavesdropping, replay, and model inversion attacks while meeting compliance standards such as GDPR and HIPAA. Results confirm that strong privacy and federated scalability can coexist under zero-trust conditions, establishing a foundation for secure cross-domain analytics in healthcare, finance, and IoT applications.
Yair Rivera Julio, Ángel D. Pinto Mangones, Nelson A. Pérez-García, Mónica-Karel Huerta · 9 authors
Large-Language-Model (LLM) functionality is rapidly becoming a cornerstone of Telemedicine-as-a-Service (PGaaS) platforms. Recent Q1 studies demonstrate that even minuscule training-set or parameter perturbations can introduce persistent back-doors, while inference pipelines leak protected health information (PHI) if left unguarded. Building on the NIST AI Risk Management Framework (AI RMF), this paper proposes and implements a zero-trust, multi-cloud security architecture that couples (i) knowledge-graph–driven data-integrity validation, (ii) containerised fine-tuning isolation, (iii) AI-RMF–centred governance and continuous risk registers, (iv) a privacy-preserving response-sanitisation gateway enhanced with one-time-password (OTP) and KYC identity binding, and (v) remote-attestation-backed zero-knowledge-proof (ZKP) integrity challenges for model weights at runtime. An extensive multi-cloud evaluation shows that the framework detects 94.6 % of tainted samples before ingestion and blocks 91.3 % of unsafe outputs, with a median latency overhead of 66 ms—well below clinical tele-consultation thresholds.
Adversarial Robustness in Machine Learning
Blockchain Technology Applications and Security
Artificial Intelligence in Healthcare and Education
Decentralized Finance (DeFi) smart contracts manage billions of dollars, making them a prime target for exploits. Price manipulation vulnerabilities, often via flash loans, are a devastating class of attacks causing significant financial losses. Existing detection methods are limited. Reactive approaches analyze attacks only after they occur, while proactive static analysis tools rely on rigid, predefined heuristics, limiting adaptability. Both depend on known attack patterns, failing to identify novel variants or comprehend complex economic logic. We propose PMDetector, a hybrid framework combining static analysis with Large Language Model (LLM)-based reasoning to proactively detect price manipulation vulnerabilities. Our approach uses a formal attack model and a three-stage pipeline. First, static taint analysis identifies potentially vulnerable code paths. Second, a two-stage LLM process filters paths by analyzing defenses and then simulates attacks to evaluate exploitability. Finally, a static analysis checker validates LLM results, retaining only high-risk paths and generating comprehensive vulnerability reports. To evaluate its effectiveness, we built a dataset of 73 real-world vulnerable and 288 benign DeFi protocols. Results show PMDetector achieves 88% precision and 90% recall with Gemini 2.5-flash, significantly outperforming state-of-the-art static analysis and LLM-based approaches. Auditing a vulnerability with PMDetector costs just $0.03 and takes 4.0 seconds with GPT-4.1, offering an efficient and cost-effective alternative to manual audits.
In edge computing, fusing privacy-sensitive heterogeneous sensor data poses challenges in balancing utility, privacy, and efficiency. Existing approaches like zkFL and zkGPT fall short in end-to-end verifiable LLM-driven fusion for non-IID data. We propose a framework embedding ZKPs into adaptive LLM layers for secure multimodal fusion with provable privacy. Key contributions: (1) context-aware attention for LLM fusion; (2) custom zk-SNARK circuits for full verification; (3) dynamic edge optimizations reducing latency by $\mathbf{2 5} \boldsymbol{\%}$. Theoretical analyses provide -DP bounds and convergence guarantees. Experiments on UCI HAR and CIFAR extensions show 91.8% accuracy, MI-AUC of 0.52, and $\mathbf{4 5 ~ m s}$ latency on Jetson Nano, outperforming zkFL by 3.5% in accuracy and 25% in efficiency.
Muhammad Tahir, Adem Orsdemir, Fiza Siyal, Uzay Işın Alıcı · 6 authors
The rapid growth in blockchain technology adoption has highlighted the significance of security in Ethereum smart contracts. Due to its immutable nature, post-deployment rectification is not possible, and vulnerabilities such as reentrancy have led to substantial financial losses in recent years, making it a pressing research priority for its timely detection. Along with static and dynamic analysis tools, recent studies have shown promising results using Deep Learning (DL) and Machine Learning (ML) techniques for vulnerability detection using imagebased methods. Although these methods often suffer from high false positive rates and limited interpretability. To address these issues, we proposed an interpretable One-dimensional Convolutional Neural Network (1D CNN), a lightweight DL framework with integrated Gradients, an attribution for the Explainable AI (XAI) framework. This framework processes smart contract opcode in a series of sequences rendered as RGB-encoded strips, enabling effective feature extraction while preserving the contract semantics and execution order. Trained on a publicly available labeled comprehensive dataset named Messi-Q, which has already been used in prominent studies in the field. Approach achieves over 97% classification accuracy in detecting reentrancy vulnerability. More importantly, it provides fine-grained, opcode-level attributions offering a scalable and interpretable path forward for smart contract analysis.
Luca Olivieri, David Beste, Luca Negrini, Lea Schönherr · 6 authors
Hyperledger Fabric (HF) is currently the one that made blockchain and smart contracts accessible to industries, providing highly customizable solutions for many enterprise use cases. Despite this, programmers are often discouraged from implementing smart contracts due to the high learning curve and security risks of naive smart contract implementations. At the same time, the advent of Large Language Models (LLMs) for code generation led to new possible scenarios such as creating new smart contract applications starting from natural language, allowing to reduce costs and development times. This paper investigates the maturity of LLMs for the code generation of HF smart contracts. In particular, we (i) generate smart contracts written in Go for HF starting from natural language descriptions, (ii) select state-of-the-art static analyzers of Go program, and (iii) perform a quality and security assessment of the generated smart contracts. Our empirical results show current LLMs do not produce high-quality smart contracts, and a relevant effort to debug and patch contracts containing bugs and possible vulnerabilities.
Hadis Rezaei, Ahmed Afif Monrat, Karl Andersson, Francesco Palmieri
The deterministic nature of blockchain technology creates fundamental difficulties in producing secure random numbers within smart contracts, a limitation that exposes vulnerabilities in applications such as decentralized finance (DeFi) protocols and blockchain-based gaming platforms. From our observations, the current state-of-the-art detection tools suffer from inadequate precision while dealing with random number vulnerabilities. To address this problem, we propose TaintSentinel, a novel path-sensitive vulnerability detection system designed to analyze smart contracts at the execution path level and gradually analyze taint with domain-specific rules. This paper discusses a solution that incorporates a multifaceted approach, integrating rule-based taint analysis to track data flow, a dual-stream neural network to identify complex vulnerability signatures, and evidence-based parameter initialization to minimize false positives. The two-phase operation of the system involves the construction of semantic graphs and the analysis of taint propagation, followed by pattern recognition using PathGNN and global structural analysis via GlobalGCN. Our experiments on 4,844 contracts demonstrate the superior performance of TaintSentinel relative to existing tools, yielding an F1-score of 0.892, an AUC-ROC of 0.94, and a PRA accuracy of 97%.
Hasan Akgul, Daniel Borg, Arta Berisha, Amina Rahimova · 6 authors
Large language models are often adapted through parameter efficient fine tuning, but current release practices provide weak assurances about what data were used and how updates were computed. We present Verifiable Fine Tuning, a protocol and system that produces succinct zero knowledge proofs that a released model was obtained from a public initialization under a declared training program and an auditable dataset commitment. The approach combines five elements. First, commitments that bind data sources, preprocessing, licenses, and per epoch quota counters to a manifest. Second, a verifiable sampler that supports public replayable and private index hiding batch selection. Third, update circuits restricted to parameter efficient fine tuning that enforce AdamW style optimizer semantics and proof friendly approximations with explicit error budgets. Fourth, recursive aggregation that folds per step proofs into per epoch and end to end certificates with millisecond verification. Fifth, provenance binding and optional trusted execution property cards that attest code identity and constants. On English and bilingual instruction mixtures, the method maintains utility within tight budgets while achieving practical proof performance. Policy quotas are enforced with zero violations, and private sampling windows show no measurable index leakage. Federated experiments demonstrate that the system composes with probabilistic audits and bandwidth constraints. These results indicate that end to end verifiable fine tuning is feasible today for real parameter efficient pipelines, closing a critical trust gap for regulated and decentralized deployments.
Machine Learning (ML) models are increasingly deployed to detect fraudulent activities in Ethereum, where phishing and scamming attacks pose serious security risks. Despite their promise, these models remain susceptible to adversarial manipulations. In this article, we present a comprehensive evaluation of ML-based Ethereum phishing detectors under a spectrum of adversarial perturbations. Our study examines multiple classifiers, including Random Forest, Decision Tree, K-Nearest Neighbors, Graph Neural Networks, and XGBoost, against rule-based, gradient-based, and black-box adversarial attacks. We conduct detailed feature-level analyses to identify transaction attributes most vulnerable to manipulation, and we evaluate the comparative robustness of classifiers under both targeted and untargeted attack scenarios. To strengthen model resilience, we assess mitigation techniques such as adversarial training and randomized smoothing, demonstrating their effectiveness in improving robustness without significant performance degradation.
Lê Thái Hùng, Huu-Han Nguyen, Thai Hung Van, Doan Minh Trung · 5 authors
Smart contract vulnerabilities pose serious risks in blockchain ecosystems, yet existing detection methods often rely on either source code or opcode analysis in isolation, missing complementary information across modalities. This paper presents a multimodal learning framework that combines semantic features extracted from source code using CodeBERT with Structure-Based Traversal (SBT) encoding and behavioral patterns derived from opcode sequences using a gMLP(gated Multi-Layer Perceptron) model applied to TF-IDF vectors. The framework systematically evaluates various fusion strategies, including concatenation, self-attention, cross-attention, and a hybrid attention mechanism, all within a unified architecture and dataset. Extensive experiments on the SmartBugs benchmark demonstrate two key findings: (1) the pairing of CodeBERT(SBT) and gMLP(opcode) achieves superior modality synergy (F1-score: 0.84), and (2) our hybrid attention fusion mechanism further improves performance to 0.87 F1, outperforming other fusion strategies by up to 3.6%. Compared to the best unimodal baselines, our approach yields a 12.8% F1 gain. To the best of our knowledge, this is the first study to provide a systematic benchmark of these fusion strategies under a unified framework for smart contract vulnerability detection. These results underscore the importance of informed modality selection and intelligent fusion design in building robust AI-driven vulnerability detection tools.
The verifiability of machine learning models and the privacy of training data have become critical concerns due to their widespread deployment in sensitive applications. Ensuring that a model performs as claimed, without revealing private data or algorithms, is a significant challenge. Zero-Knowledge Proof systems (ZKPs) have emerged as a promising cryptographic solution, enabling the verification of statements without disclosing underlying information. Their integration with blockchain technology further enhances trust and decentralization, offering robust solutions for secure and transparent AI systems. This paper explores the use of ZKPs in machine learning, focusing on privacy-preservation techniques, model verifiability, and confidential AI. It compares the differences and challenges of employing ZKPs in machine learning versus blockchains, highlighting their unique requirements and overlapping benefits. We review the basic concepts of ZKPs, advances such as zkSNARKs and zk-STARKs, and their applications in blockchainbased AI frameworks to ensure data integrity, immutability, and scalability. Furthermore, the paper delves into the practical implications of using ZKPs in AI, providing case studies and analyzing their scalability, performance, and limitations. We conclude by identifying key challenges and presenting future research directions to extend the applicability of ZKPs in AI, particularly in federated learning, model fairness, and decentralized AI pipelines.
Mohammad Alja’afreh, Sarah Tarawneh, Hikmat Adhami, Ali Karime · 5 authors
The metaverse—a persistent, multiuser fusion of digitally augmented reality and computer-generated virtuality— is emerging as a programmable substrate for identity, assets, and interaction. Its heterogeneous stack (XR clients, engines/SDKs, Web3 rails, wallets, marketplaces) enlarges the attack surface. This paper contributes: (i) a structured threat taxonomy specialized for Web3/XR platforms; (ii) explicit system and adversary models; (iii) a risk quantification scheme combining behavioral and on-chain signals; and (iv) a data-driven defense architecture aligning decentralized identity, wallet/custody guardrails, analytics, AI-aided detection, and policy instrumentation. We further instantiate these controls in the Medical MeTAI context, where confidentiality, integrity, and provenance requirements are stringent.
We present a cross-market algorithmic trading system that balances execution quality with rigorous compliance enforcement. The architecture comprises a high-level planner, a reinforcement learning execution agent, and an independent compliance agent. We formulate trade execution as a constrained Markov decision process with hard constraints on participation limits, price bands, and self-trading avoidance. The execution agent is trained with proximal policy optimization, while a runtime action-shield projects any unsafe action into a feasible set. To support auditability without exposing proprietary signals, we add a zero-knowledge compliance audit layer that produces cryptographic proofs that all actions satisfied the constraints. We evaluate in a multi-venue, ABIDES-based simulator and compare against standard baselines (e.g., TWAP, VWAP). The learned policy reduces implementation shortfall and variance while exhibiting no observed constraint violations across stress scenarios including elevated latency, partial fills, compliance module toggling, and varying constraint limits. We report effects at the 95% confidence level using paired t-tests and examine tail risk via CVaR. We situate the work at the intersection of optimal execution, safe reinforcement learning, regulatory technology, and verifiable AI, and discuss ethical considerations, limitations (e.g., modeling assumptions and computational overhead), and paths to real-world deployment.
A. G. Ramakrishnan, Shubham Agarwal, Sharmila Kumari Selvanayagam, Kunwar P. Singh
As image generation models grow increasingly powerful and accessible, concerns around authenticity, ownership, and misuse of synthetic media have become critical. The ability to generate lifelike images indistinguishable from real ones introduces risks such as misinformation, deepfakes, and intellectual property violations. Traditional watermarking methods either degrade image quality, are easily removed, or require access to confidential model internals – making them unsuitable for secure and scalable deployment. We are the first to introduce ZK-WAGON, a novel system for watermarking image generation models using the Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (ZK-SNARKs). Our approach enables verifiable proof of origin without exposing model weights, generation prompts, or any sensitive internal information. We propose Selective Layer ZK-Circuit Creation (SL-ZKCC), a method to selectively convert key layers of an image generation model into a circuit, reducing proof generation time significantly. Generated ZK-SNARK proofs are imperceptibly embedded into a generated image via Least Significant Bit (LSB) steganography. We demonstrate this system on both GAN and Diffusion models, providing a secure, model-agnostic pipeline for trustworthy AI image generation.
Open access
2 source records
Physical Unclonable Functions (PUFs) and Hardware Security
Adversarial Robustness in Machine Learning
Generative Adversarial Networks and Image Synthesis
This paper addresses the critical systemic risk of AI Safety Arbitrage, where users exploit inconsistent safety standards across jurisdictions to access restricted capabilities. Through a controlled red-team test, we demonstrate how current frameworks fail to prevent the extraction of hazardous procedural knowledge, leaving these failures unreported and without legal consequence. To resolve this, we propose a Global Socio-Technical Architecture for AI Accountability based on distributed ledger technology (DLT). This infrastructure creates a protocol network that is conceptually similar to TCP/IP but for accountability designed to align incentives through transparency and cryptographic verification. Key Contributions & ArchitectureThe proposed solution rests on four pillars designed to replace trust relationships with cryptographic verification: Globally Unique Model Registration: Establishes digital identities (DIDs) for AI systems with value chain provenance. Independent Auditor Certification: Licensed validators stake economic value on certification accuracy, removing the need to trust model provider claims. Hardware-Backed Attestation: Tamper-resistant verification ensures deployed systems adhere to registered specifications. Continuous Reputation Monitoring: Oracle networks provide ongoing assessment of compliance with automated penalties for fraud. Technical & Governance Implementation Zero-Knowledge Proofs (ZKP): We illustrate technical viability using zkEVM technology. This allows auditors to prove compliance with safety standards without revealing proprietary training data or model architectures, resolving the tension between accountability and Intellectual Property protection. The AIAO Framework: Inspired by the International Civil Aviation Organization (ICAO), we propose the AI Accountability Coordination Organization (AIAO). This body defines "red-line" safety primitives that nations voluntarily adopt, allowing for regulatory sovereignty while ensuring global interoperability. ConclusionBy breaking the "regulatory arbitrage cycle," this framework enables a transition from safety theater to verifiable safety. It supports open-source innovation through graduated oversight and reputation systems, ensuring that AI development remains both agile and accountable.
Artificial intelligence (AI) is increasingly central to solving complex societal, economic, and scientific problems, yet prevailing models remain constrained by their opacity, vulnerability to adversarial inputs, and reliance on centralized infrastructures. These limitations underscore the urgent need for approaches that combine the adaptability of neural networks with the interpretability and rule-based precision of symbolic systems. At the same time, decentralization has emerged as a critical paradigm for enhancing trust, resilience, and accountability in intelligent systems. Together, these threads converge on the concept of decentralized neuro-symbolic cognitive systems, which integrate distributed inference, symbolic reasoning, and governance mechanisms to create secure and transparent frameworks for machine intelligence. This article presents a comprehensive methodology for the design and operation of such systems, advancing beyond conventional hybrid AI by embedding causal intent routing, federated cognitive capsules, encrypted episodic memory, and immutable epistemic ledgers. These elements are supported by governance innovations such as the NeuroConstitution™, which enables tokenized, evolvable norms and ensures accountability through transparent dispute resolution. The framework is evaluated across key application domains, including healthcare, finance, governance, and climate modeling, with comparative benchmarks demonstrating gains in robustness, interpretability, and systemic trust. By uniting symbolic reasoning, neural inference, and decentralized governance, this research outlines a pathway toward AI systems that are not only technically powerful but also socially aligned and ethically defensible. The article concludes that decentralized neuro-symbolic architectures provide a sustainable foundation for advancing trustworthy AI capable of supporting critical infrastructures and decision-making in a rapidly evolving world.
In the Internet of Vehicles (IoV) network, numerous vehicle terminals are required to continuously upload local data to maintain the latest service models, which supports intelligent transportation and personalized services. However, the privacy risks posed by this continuous data uploading cannot be ignored. Federated learning, as a distributed ma-chine learning paradigm, enables global model training without sharing original data. The introduction of blockchain further supports decentralization and immutability. However, federated learning also faces the risk of poisoning attacks, where malicious clients may upload abnormal or tampered model updates, severely impacting global model performance. To address this, this paper proposes a security framework that combines zero-knowledge proofs, federated learning, and blockchain. Clients use zero-knowledge proofs to ensure the legitimacy of uploaded updates, while the blockchain is responsible for verification and storage. Ultimately, a robust global model is obtained through federated aggregation. Experimental results demonstrate that this scheme effectively resists poisoning attacks, significantly improving system security and reliability while protecting user privacy.