LLMs for security
Finding vulnerabilities, repairing code, and checking that generated software behaves as intended.
Explore LLM4SECTrust AI &
Security Lab
Sungshin Women's University
Department of Convergence Security Engineering
We study how language models can secure software—and how to secure AI models and agents themselves.
Security with AI.
Security for AI.
Two connected directions, grounded in software and system evaluation.
Security for AI systems
Examining RAG attacks, agent assurance, and privacy and integrity risks in federated learning.
Explore SEC4LLMSelected research
All publicationsExecution-based evaluation of fine-tuned language models for automated vulnerability repair.
Read the paperFinding vulnerabilities with local context and verification
Empirical Software Engineering · Accepted 2026
VulScope studies software vulnerability detection through local code-context analysis and verification.
Read the paperUnderstanding privacy risks in federated learning
IEEE Access · 2026
PI-FD examines property inference against federated learning with distillation using synthetic data generation.
Read the paperThe people behind the research
Younghan Lee
Assistant Professor, Sungshin Women's University
Our undergraduate researchers work on ROS 2 software evaluation and RAG security.
Meet the teamNews
All updates| Nov 14, 2026 | TAIS Lab has moved to taislab.com. Our research, publications, and team are available at the new address. |
|---|---|
| Jun 06, 2026 | “PI-FD: Property Inference Attacks on Federated Learning With Distillation via Synthetic Data Generation” was published in IEEE Access. |
| May 11, 2026 | “VulScope: Software Vulnerability Detection via Local Context Analysis and Verification” was accepted for publication in Empirical Software Engineering. |
| Apr 15, 2026 | Our study, “Rethinking the Capability of Fine-Tuned Language Models for Automated Vulnerability Repair,” was presented in the ICSE 2026 Research Track. The work introduces execution-based evaluation for learning-based vulnerability repair. |
Let's talk about your research interests.
For student enquiries, research collaboration, or questions about our work, get in touch.