Abstract Smart contract vulnerabilities have caused billions of dollars in losses across decentralized finance. Finding reliable ways to detect such vulnerabilities has been a long-standing challenge for researchers. The growing capabilities of large language models (LLMs) are promising, but the factors that determine their reliability and capabilities remain poorly understood. This study investigates whether increasing inference-time computation using techniques like extended reasoning and structured prompting always improves vulnerability detection capability. It also identifies the most influential factors to select a model for this task. Using four prompting techniques, it evaluates 14 LLMs from seven families on 54 Solidity contracts. The experiment reveals a clear gap in detection capability across model classes. While six frontier models do not report false positives on verified-clean contracts, all three small open-source models report vulnerabilities in every case throughout the experiment. Moreover, a 11.5% drop in F1 score for one model was observed when increasing inference-time compute by enabling extended thinking. Also, prompting strategy has a limited effect on detection capability compared to model selection. The results challenge common assumptions and offer practical insights into the use of LLMs for smart contract vulnerability detection.
The rapid digitization of financial services has created increasingly complex environments in which financial institutions must process large volumes of heterogeneous transaction data while simultaneously protecting customers, detecting fraud, managing financial risks, and complying with regulatory requirements. Traditional centralized and rule-based fraud detection systems face significant challenges associated with data volume, processing latency, evolving fraudulent behaviors, class imbalance, and the increasing sophistication of cyber-enabled financial crimes. This paper proposes a distributed intelligent analytics framework for blockchain-based fraud detection and risk management in financial institutions. The framework integrates distributed big data analytics, artificial intelligence, machine learning, blockchain, graph-based learning, and intelligent decision support into a unified architecture. Distributed computing provides scalable processing of heterogeneous financial datasets, while artificial intelligence identifies anomalous transactions and predicts potential risks. Blockchain provides a complementary integrity, traceability, and verification layer for financial transactions. Graph Neural Networks can further model relationships among customers, accounts, devices, merchants, and transactions, enabling the detection of complex fraud patterns that may not be visible through transaction-level analysis. The framework builds on Ramareddy's work on distributed big data analytics for scalable knowledge discovery in heterogeneous systems and Chhunchha's investigation of blockchain's influence on financial institutions. Recent research also indicates growing interest in machine learning, graph-based models, federated learning, and blockchain for financial fraud detection. The proposed framework addresses important challenges including scalability, privacy, class imbalance, concept drift, explainability, cybersecurity, and regulatory compliance. The paper argues that combining distributed analytics with blockchain and AI can provide financial institutions with a more scalable, transparent, adaptive, and intelligent approach to fraud prevention and financial risk management.
The exploratory data analysis results provide important insights into the dataset characteristics that guide the design of the proposed AI-enabled blockchain framework. The class distribution graph shows a strong imbalance, with approximately 86.2% genuine samples and 13.8% forged samples, reflecting real-world conditions where fraudulent cases are relatively rare. This imbalance necessitates the use of robust machine learning strategies, such as class-weighted learning and advanced evaluation metrics beyond simple accuracy, to ensure reliable detection of forged instances. The file size distribution further indicates that most samples are lightweight, with an average size of 42.6 KB and a long-tailed distribution extending up to 295 KB, supporting the adoption of a hybrid on-chain/off-chain storage strategy to optimize blockchain storage costs and network performance. Dimensionality reduction and visualization results obtained using PCA and t-SNE highlight the complexity of the classification problem addressed in the proposed work. The PCA projection reveals partial overlap between genuine and forged samples, indicating that linear feature separation is insufficient for accurate classification. Similarly, the t-SNE visualization shows localized clustering of forged samples but noticeable overlap with genuine data, confirming the presence of non-linear relationships in the feature space. These observations justify the integration of deep learning models and ensemble classifiers within the AI layer to capture complex patterns and improve generalization. The image resolution distribution further demonstrates that most images fall within a consistent resolution range of approximately 300–700 pixels (width) and 200–550 pixels (height), ensuring stable model training while still requiring standardized preprocessing to handle resolution variability across training, validation, and test splits. Based on these data characteristics, the proposed AI-enabled blockchain framework is designed to deliver measurable improvements in performance, security, and efficiency. Experimental evaluation shows that the AI-driven valuation and classification modules achieve a fraud detection accuracy of 94.1%, with a precision of 91.6%, recall of 89.3%, and an F1-score of 90.4%, demonstrating reliable performance despite class imbalance. The blockchain layer achieves an average throughput of approximately 420 transactions per second with a confirmation latency of 2.6 seconds, while maintaining a low transaction cost of ₹18–₹25 per transaction through Layer-2 scaling and off-chain storage optimization. Smart contracts exhibit a 99.1% execution success rate and high vulnerability detection coverage during security analysis, validating the robustness of automated transaction execution. The expected outcomes of the proposed system include reduced transaction settlement time, enhanced fraud resistance, improved valuation transparency, and greater market accessibility through tokenization and fractional ownership. By combining AI-driven intelligence with blockchain-based trust and automation, the framework is expected to significantly reduce manual intervention, operational costs, and regulatory non-compliance risks in real estate transactions. Overall, the results and projections confirm that the proposed approach is well-suited for real-world deployment, offering a scalable, secure, and intelligent solution for next-generation real estate asset management systems.
In blockchain-enabled supply chain finance, traditional credit risk assessment models suffer from conflicts between data sharing and privacy protection, reliance on static evaluation methods, and limited data credibility. To overcome these challenges, this paper proposes a blockchain-based dynamic credit risk assessment model that integrates privacy computing and intelligent risk monitoring. First, blockchain’s immutability and traceability ensure the authenticity and transparency of supply chain transaction data, effectively mitigating information asymmetry and data tampering. Second, privacy-preserving technologies, including homomorphic encryption based on the Paillier algorithm and zk-SNARKs, enable secure data sharing and validity verification without exposing sensitive enterprise information, thereby improving assessment reliability. Third, a dynamic risk monitoring framework is constructed by combining smart contracts, long short-term memory (LSTM) networks, and an improved dynamic graph neural network (DGNN). LSTM models temporal risk evolution in transaction data, while DGNN captures risk propagation among upstream and downstream enterprises. Smart contracts synchronize transaction states in real time, allowing continuous updates of credit risk levels. The proposed secure information processing and dynamic graph modeling strategy also provides a valuable reference for trustworthy data interaction and intelligent decision-making in distributed electromagnetic sensing and communication networks, where reliable information propagation and adaptive resource management are essential. Experimental results based on a textile supply chain dataset show that the proposed model achieves approximately 94% credit assessment accuracy, outperforming traditional static models by 15%–20%, while maintaining excellent response speed and throughput for dynamic financial decision-making. The proposed framework provides a practical and secure solution for blockchain-based credit risk management and offers methodological insights for data-driven engineering systems requiring secure information fusion and dynamic network analysis.
The increasing demand for trustworthy and privacy-preserving credit reporting systems has exposed the limitations of both centralized and existing blockchain-based solutions, including scalability bottlenecks, weak privacy protection, and insufficient incentive mechanisms. To address these challenges, we propose LightCred, a novel consortium blockchain-based personal credit management framework that integrates lightweight nodes, Merkle proofs, multi-role smart contracts, and privacy-preserving cryptographic techniques. LightCred features a five-layer architecture that efficiently collects, verifies, stores, and serves credit data while ensuring data integrity, confidentiality, and regulatory compliance. Specifically, it (i) employs a low-cost and traceable data reduction mechanism through lightweight nodes and Merkle proofs to minimize storage and improve verifiability; (ii) introduces a multi-role smart contract model that enforces dynamic access control and fair incentive distribution based on participant reputations; and (iii) integrates zero-knowledge proofs and homomorphic encryption to support privacy-preserving credit scoring and querying. Experimental results demonstrate that LightCred achieves superior performance compared to five baseline methods, delivering up to 5% higher throughput, 3–5% lower privacy leakage, and 10–15% reduced storage costs, while maintaining competitive latency and auditability. These findings validate LightCred as a robust, scalable, and privacy-aware credit management solution, offering a viable alternative for modern credit reporting systems.