About Me

Hi! I am Geon Woo Kim, an undergraduate researcher at Sungkyunkwan University, pursuing a double major in Computer Education and Artificial Intelligence. Since September 2026, I have also been enrolled in the Bachelor’s-Master’s Integrated Program in Artificial Intelligence. I am a research student at SWLab, advised by Prof. Moohong Min.

My research interests include Trustworthy AI and Watermarked Knowledge Bases for RAG. My research experience also includes internships at the Korea AI Safety Institute and ETRI.

🔥News

• 2026.10: 🏆 Received the Excellence Award at the AI Forensic Investigation Challenge 2026, ranking 4th out of 174 teams (Top 2.3%)!
• 2026.09: 🎉 Our paper “Don't Just Impute, Detect: Task-Aware Diffusion Pipelines for Edge Anomaly Detection” was accepted to HiPC 2026! See you in Bengaluru, India!
• 2026.07 - 2026.08: Research internship at the Korea AI Safety Institute (AISI), researching PII detection and context-/task-aware approaches for foundation models.

🔍Research Interests

• Trustworthy AI
• Watermarked Knowledge Bases for RAG

📖Education

📝Publications

HiPC 2026

Don't Just Impute, Detect: Task-Aware Diffusion Pipelines for Edge Anomaly Detection

Jeongsu Park, Taehyeon Yun, Geon Woo Kim, Yuji Lim, Moohong Min*.
33rd IEEE International Conference on High Performance Computing, Data, and Analytics (HiPC 2026, BK=1)

ETRI Journal

Lost in the Comments: Social Context as a Single-Pass Jailbreak and Defense on Agentic Platforms

Taehyeon Yun, Seonhee Jeon, Geon Woo Kim, Jeongsu Park, Moohong Min*.
ETRI Journal (SCIE)

Link

ETRI Journal

Data Poisoning in LLM Multi-Agent Societies: Social Proof Drives Collective Decision Failure in Financial Deliberation

Jeongsu Park, Yuji Lim, Geon Woo Kim, Taehyeon Yun, Moohong Min*.
ETRI Journal (SCIE)

Link

IEEE Access

A Taxonomy of Knowledge Bases for Retrieval-Augmented Methods in Vision: A Comprehensive Survey

Geon Woo Kim, Dong Hwan Lee, Janghee Yoo*.
IEEE Access (SCIE)

Link

CISC-S’25

A Pipeline for Automatically Extracting Illegal Gambling Links from Spam Messages

Geon Woo Kim, Jung Hyun Chun, Moohong Min*.
CISC-S'25 (Poster, Domestic)

PDF

📄Manuscripts Under Review

Manuscripts currently under review. Public links are provided only when available.

Under Review

SONAR: Keyed Retrieval Signatures for Search-Only Auditing of Image Indexes

Geon Woo Kim, Minseok Hur, Moohong Min*. Under Review

[GitHub]

Under Review

KNOT: Watermarking RAG Knowledge Bases for Ownership Verification via Multi-hop Traces

Minseok Hur, Geon Woo Kim, Moohong Min*. Under Review

[GitHub]

Under Review

Search Well, Share Less: Retrieval-Aware Query Rewriting for Private Queries

Geon Woo Kim, Taehyeon Yun, Moohong Min*. Under Review

💻Research & Work Experience

🛠Projects

• 2026.01 - Present, Attack Scenario Analysis and Hybrid RAG Knowledge Base Construction Based on MITRE ATT&CK and D3FEND, part of the Development of an AI White-Hacker-based Zero-Hacking System program led by the Korea Institute of Science and Technology Information (KISTI), in collaboration with KAIST.

  Designing and building the ATT&CK/D3FEND-grounded knowledge database that enables an LLM-based attack profiler to reason over multi-stage intrusion scenarios.

• 2025.03 - Present, Development of Cyber Security Regulation Technologies for Programmable Logic Devices in the Development Phase of New Nuclear Reactors, funded by the Korea Foundation of Nuclear Safety (KoFONS).

  Analyzed the FPGA development lifecycle, existing nuclear safety regulatory requirements, and known security vulnerabilities to derive development-phase cybersecurity requirements.

• 2026.07 - 2026.08, Development of Privacy Risk Management Technologies for Training Data of Foundation Models, supported by the Personal Information Protection Commission (PIPC), administered by the Korea Internet & Security Agency (KISA), and led by the Korea AI Safety Institute (AISI), ETRI.

  Reviewed literature on context-aware personally identifiable information (PII) detection and designed and implemented an automated framework.

• 2025.07 - 2025.08, Development of AI Technology for Early Screening of Infant/Child Autism Spectrum Disorders Based on Cognition of Psychological Behavior and Response, funded by the Ministry of Science and ICT (MSIT) through an Institute of Information & Communications Technology Planning & Evaluation (IITP) grant.

  Analyzed cognitive behavioral cues and responses to support AI-based early screening of infant/child autism spectrum disorders.

🏅Honors & Awards

📚Others

🌎Visitor Map