AIS2 Lab
Lingnan University, Hong Kong SAR.
Welcome to AIS2 (Artificial Intelligence and Systems Security) Lab! AIS2Lab is an AI-oriented security research team established by Asst Prof. Daoyuan Wu in the Division of Industrial Data Science, School of Data Science at Lingnan University (LU), one of the eight UGC-funded universities in Hong Kong.
We adopt a systems security mindset to advance the security and trustworthiness of artificial intelligence in the era of Large Language Models (LLMs). At AIS2Lab, we are committed to conducting inter-disciplinary research that integrates knowledge and methodologies from computer science, artificial intelligence, law, healthcare, and other related fields. Our work aims to address emerging security challenges and promote responsible innovation across both technical and societal domains.
Specifically, our research focuses on the following key areas:
Large Language Model and AI Security: LLMs for Cybersecurity; Security of LLMs; AI Safety; LLM + Law.
Blockchain and Smart Contract Security: Chain & DeFi Security; Consensus Security; Transaction Compliance.
GPU Software and Medical System Security: AI Infrastructure Security; Healthcare and Medical System Security.
Novel Program Analysis and Mobile Security: Novel Program Analysis & Fuzzing; LLM for Mobile; EdgeAI Security.
Through close collaboration with experts from diverse disciplines, AIS2Lab aims to build secure and responsible AI systems that benefit both technology and society. To realize this vision, we are always seeking passionate and persistent students (PhD/RA/Postdoc/Interns) with backgrounds or strong interests in AI/LLMs, blockchain, GPU and medical software, programming languages, and fuzzing to join AIS2Lab. We value persistence and a commitment to research excellence.
Currently, we have the following priority openings:
Self-financed PhD/MPhil positions are always available. They are used only in cases where applicants do not meet the scholarship requirements set by the university committee but are still academically qualified to pursue PhD/MPhil studies. In such cases, I will support you as a part-time RA, which will cover your tuition fees and provide some living allowance.
news
| Jan 01, 2026 | Our lab page goes online! |
|---|---|
| Sep 26, 2025 | Four papers accepted by IEEE/ACM ASE 2025. |
| Aug 16, 2025 | PI joined Lingnan University as a tenure-track Assistant Professor. |
selected publications
- ASEDetecting Various DeFi Price Manipulations with LLM ReasoningIn Proc. IEEE/ACM Automated Software Engineering (ASE), 2025
- ASEDemystifying OpenZeppelin’s Own Vulnerabilities and Analyzing Their Propagation in Smart ContractsIn Proc. IEEE/ACM Automated Software Engineering (ASE), 2025
- ASEHave We Solved Access Control Vulnerability Detection in Smart Contracts? A Benchmark StudyIn Proc. IEEE/ACM Automated Software Engineering (ASE), 2025
- TSEACFix: Guiding LLMs with Mined Common RBAC Practices for Context-Aware Repair of Access Control Vulnerabilities in Smart ContractsIEEE Transactions on Software Engineering (TSE), 2025
- CCSMeasuring and Augmenting Large Language Models for Solving Capture-the-Flag ChallengesIn Proc. ACM SIGSAC Conference on Computer and Communications Security (CCS), 2025
- CCSDifferentiation-Based Extraction of Proprietary Data from Fine-tuned LLMsIn Proc. ACM SIGSAC Conference on Computer and Communications Security (CCS), 2025
- OOPSLAAPI-guided Dataset Synthesis to Finetune Large Code ModelsProceedings of the ACM on Programming Languages (OOPSLA), 2025
- USENIX SecuritySelfDefend: LLMs Can Defend Themselves against Jailbreaking in a Practical MannerIn Proc. USENIX Security Symposium, 2025
- USENIX SecurityLow-Cost and Comprehensive Non-textual Input Fuzzing with LLM-Synthesized Input GeneratorsIn Proc. USENIX Security Symposium, 2025
- ISSTADecLLM: LLM-Augmented Recompilable Decompilation for Enabling Programmatic Use of Decompiled CodeProceedings of the ACM on Software Engineering (ISSTA), 2025
- ICSETesting and Understanding Deviation Behaviors in FHE-hardened Machine Learning ModelsIn Proc. IEEE/ACM Conference on Software Engineering (ICSE), 2025
- ICSECombining Fine-Tuning and LLM-Based Agents for Intuitive Smart Contract Auditing with JustificationsIn Proc. IEEE/ACM Conference on Software Engineering (ICSE), 2025
- NDSSPropertyGPT: LLM-driven Formal Verification of Smart Contracts through Retrieval-Augmented Property GenerationIn Proc. ISOC Network and Distributed System Security Symposium (NDSS), 2025
- DLTAGChain: A Blockchain-based Gateway for Trustworthy App Delegation from Mobile App MarketsACM Distributed Ledger Technologies: Research and Practice (DLT), 2024
- EMNLPSplit and Merge: Aligning Position Biases in LLM-based EvaluatorsIn Proc. Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024
- USENIX SecurityUsing My Functions Should Follow My Checks: Understanding and Detecting Insecure OpenZeppelin Code in Smart ContractsIn Proc. USENIX Security Symposium, 2024
- EuroS&PMtdScout: Complementing the Identification of Insecure Methods in Android Apps via Source-to-Bytecode Signature Generation and Tree-based Layered SearchIn Proc. IEEE European Symposium on Security and Privacy (EuroS&P), 2024
- ICSEGPTScan: Detecting Logic Vulnerabilities in Smart Contracts by Combining GPT with Program AnalysisIn Proc. IEEE/ACM Conference on Software Engineering (ICSE), 2024
- ISSTABeyond “Protected” and “Private”: An Empirical Security Analysis of Custom Function Modifiers in Smart ContractsIn Proc. ACM International Symposium on Software Testing and Analysis (ISSTA), 2023
- NDSSBlockScope: Detecting and Investigating Propagated Vulnerabilities in Forked Blockchain ProjectsIn Proc. ISOC Network and Distributed System Security Symposium (NDSS), 2023
- FSEAn Empirical Study of Blockchain System Vulnerabilities: Modules, Types, and PatternsIn Proc. ACM Symposium on the Foundations of Software Engineering (FSE), 2022
- RAIDOn the Usability (In)Security of In-App Browsing Interfaces in Mobile AppsIn Proc. International Symposium on Research in Attacks, Intrusions and Defenses (RAID), 2021
- DSNWhen Program Analysis Meets Bytecode Search: Targeted and Efficient Inter-procedural Analysis of Modern Android Apps in BackDroidIn Proc. IEEE/IFIP International Conference on Dependable Systems and Networks (DSN), 2021
- EMSEScalable Online Vetting of Android Apps for Measuring Declared SDK Versions and Their Consistency with API CallsSpringer Empirical Software Engineering (EMSE), 2021
- DIMVAUnderstanding Android VoIP Security: A System-Level Vulnerability AssessmentIn Springer International Conference on Detection of Intrusions and Malware, and Vulnerability Assessment (DIMVA), 2020
- AsiaCCSTowards Understanding Android System Vulnerabilities: Techniques and InsightsIn Proc. ACM Asia Conference on Computer and Communications Security (AsiaCCS), 2019
- NDSSUnderstanding Open Ports in Android Applications: Discovery, Diagnosis, and Security AssessmentIn Proc. ISOC Network and Distributed System Security Symposium (NDSS), 2019
- CODASPYSCLib: A Practical and Lightweight Defense against Component Hijacking in Android ApplicationsIn Proc. ACM Conference on Data and Applications Security and Privacy (CODASPY), 2018
- USENIX ATCMopEye: Opportunistic Monitoring of Per-app Mobile Network PerformanceIn Proc. USENIX Annual Technical Conference (ATC), 2017
- WASAMeasuring the Declared SDK Versions and Their Consistency with API Calls in Android AppsIn Proc. Springer International Conference on Wireless Algorithms, Systems, and Applications (WASA), 2017
- MoSTIndirect File Leaks in Mobile ApplicationsIn Proc. IEEE Mobile Security Technologies (MoST), in conjunction with S&P, 2015
- ISCAnalyzing Android Browser Apps for file:// VulnerabilitiesIn Proc. Springer Information Security Conference (ISC), 2014