Curriculum Vitae

Minkyoo Song

Ph.D. Student in Electrical Engineering at the Network and System Security Lab, KAIST. Advisor: Prof. Seungwon Shin.

I work on LLM security, AI for security, data-driven security, and social network analysis, with a focus on understanding vulnerabilities in modern AI systems and building defenses for practical deployment.

CV PDF Mail LinkedIn Google Scholar

Last update: January 2026

Research Interest

Research areas

  • LLM Security
  • AI for Security
  • Data-driven Security
  • Social Network Analysis

Education

Academic background

Korea Advanced Institute of Science and Technology (KAIST)

March 2023 - February 2027 (estimated)

Ph.D. Student in Electrical Engineering, Network and System Security Lab. Advisor: Seungwon Shin.

Daejeon, South Korea

Korea Advanced Institute of Science and Technology (KAIST)

March 2021 - February 2023

M.S. in Electrical Engineering, Network and System Security Lab. Advisor: Seungwon Shin.

Daejeon, South Korea

Korea Advanced Institute of Science and Technology (KAIST)

March 2016 - February 2021

B.S. in Industrial and Systems Engineering, double majored in Electrical Engineering.

Daejeon, South Korea

Publications

Publications

[C]: Conference, [J]: Journal, [U]: Under Review

  1. [C]

    W. Choi*, M. Seo*, M. Song, H. Heo, S. Shin, M. You.

    PC^2: Politically Controversial Content Generation via Jailbreaking Attacks on GPT-based Text-to-Image Models

    33rd ACM Conference on Computer and Communications Security (CCS 2026)

  2. [C]

    J. Kim, M. Song, S. Shin, S. Son.

    SafeMoE: Safe Fine-Tuning for MoE LLMs by Aligning Harmful Input Routing

    The Fourteenth International Conference on Learning Representations (ICLR 2026)

  3. [C]

    J. Kim, S.H. Na, M. Song, S. Shin, S. Son.

    MoEvil: Poisoning Expert to Compromise the Safety of Mixture-of-Experts LLMs

    2025 Annual Computer Security Applications Conference (ACSAC 2025). Distinguished Paper Award.

  4. [C]

    M. Song, H. Kim, J. Kim, S. Shin, S. Son.

    Refusal Is Not an Option: Unlearning Safety Alignment of Large Language Models

    34th USENIX Security Symposium (SECURITY 2025)

  5. [C]

    H. Kim, M. Song, S.H. Na, S. Shin, K. Lee.

    When LLMs Go Online: The Emerging Threat of Web-Enabled LLMs

    34th USENIX Security Symposium (SECURITY 2025)

  6. [C]

    M. Song, H. Kim, J. Kim, Y. Jin, S. Shin.

    Claim-Guided Textual Backdoor Attack for Practical Applications

    The 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025 Findings)

  7. [C]

    J. Kim, M. Song, S.H. Na, S. Shin.

    Obliviate: Neutralizing Task-Agnostic Backdoors within the Parameter-Efficient Fine-Tuning Paradigm

    The 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025 Findings)

  8. [C]

    M. Song, E. Jang, J. Kim, S. Shin.

    Covering Cracks in Content Moderation: Delexicalized Distant Supervision for Illicit Drug Jargon Detection

    31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025)

  9. [C]

    J. Kim, M. Song, M. Seo, Y. Jin, S. Shin.

    PassREfinder: Credential Stuffing Risk Prediction by Representing Password Reuse between Websites on a Graph

    2024 IEEE Symposium on Security and Privacy (Oakland 2024)

  10. [J]
  11. [J]

    J. Choi, J. Kim, M. Song, H. Kim, N. Park, M. Seo, Y. Jin, S. Shin.

    A Large-Scale Bitcoin Abuse Measurement and Clustering Analysis Utilizing Public Reports

    IEICE Transactions on Information and Systems

  12. [U]

    M. Song, J. Kim, M. Kang, H. Kim, S. Shin, S. Son.

    Graphs Don't Stay Secret: Practical Subgraph Reconstruction Attacks on Defended Graph RAG

    Under Review

  13. [U]
  14. [U]

    K. Kim, J. Cui, M. Song, S. Shin.

    Exploring the Familiar Taste of Toxicity: A Causal Influence Analysis of Toxic Comments on Internet Forums

    Invited to Major Revision at IEEE Transactions on Knowledge and Data Engineering (TKDE)

  15. [U]
  16. [U]
  17. [U]

Experience

Professional experience

S2W

July 2022 - Feb 2023

Research Intern @ AI Team, South Korea

  • Illicit drug jargon detection: analyzed illicit drug-related discussions and developed an LLM-based content moderation framework, independently capturing contextual and lexical characteristics.

Honors and Awards

Recognition

Distinguished Paper Award

Annual Computer Security Applications Conference (ACSAC)

MoEvil: Poisoning Expert to Compromise the Safety of Mixture-of-Experts LLMs

2025

4th Prize, 2025 Cybersecurity Paper Competition

Korean Association of Cybersecurity Studies (KACS)

Poisoning Expert to Compromise the Safety of Mixture-of-Experts LLMs

2025

2nd Prize, 2023 Cybersecurity Paper Competition

Korean Association of Cybersecurity Studies (KACS)

Graph-based Deep Learning Framework for Credential Stuffing Risk Prediction

2023

4th Prize, 2023 Cybersecurity Paper Competition

Korean Association of Cybersecurity Studies (KACS)

Delexicalized Distant Supervision for Illicit Drug Jargon Detection

2023

4th Prize, 2023 Cybersecurity Paper Competition

Korean Association of Cybersecurity Studies (KACS)

Understanding the Occurrence and Impact of Credential Data Breach

2023

Cum Laude

Korea Advanced Institute of Science and Technology (KAIST)

2021

Academic Achievement Award: Salutatorian

Korea Advanced Institute of Science and Technology (KAIST)

2019 Spring

Dean's List

Industrial and Systems Engineering (ISysE, KAIST)

2019 Spring

Languages

Languages

  • KoreanNative
  • EnglishFluent

Contact

For collaborations or inquiries, please get in touch.

MailLinkedInGoogle Scholar