AI safety & speech security

Seungmin Kim 김승민

Research Intern · AIM Intelligence

M.S. Student · Soongsil University

I am an M.S. student in AI Convergence Security at Soongsil University and a research intern at AIM Intelligence, working on AI safety — adversarial robustness and the security of speech AI. My research centers on voice protection against voice-cloning attacks, and more recently on the safety and security of Audio Language Models (ALMs).

  • AI safety
  • Adversarial robustness
  • Speech AI security
Portrait of Seungmin Kim

Experience

Sep. 2026 – Present

Research Intern

AIM Intelligence

Jul. 2022 – Feb. 2025

Undergraduate Student Intern

AI Safety Center, Soongsil University

Speech AI security research — deepfake-voice detection and proactive voice protection against AI-based voice cloning.

Education

2025 – 2027

M.S. in AI Convergence Security, Soongsil University

Expected Feb. 2027 · GPA 4.5 / 4.5

Research focus: AI safety, adversarial robustness, speech AI security.

2021 – 2025

B.S. in Software, Soongsil University

GPA 3.0 / 4.5

News

Recent updates

Selected publications

All publications
NeurIPS 2026Accepted2026

MUTE: Multi-Level Alignment Uncoupling Against Talking-Head Exploitation for Voice Protection

Donghyun Kim*, Jin Hong*, Seungmin Kim*, Dain Kim, Junseok Kwon, Daeseon Choi

*These authors contributed equally and are listed in alphabetical order.

NaVo generation and voice protection framework
Interspeech 2026Sep. 2026

NaVo: Natural Voice Protection against Voice Cloning Attacks via Generative Universal Adversarial Audio

Seoyoung Park*, Seungmin Kim*, Sohee Park, Dain Kim, Thien An Nguyen, Thien-Phuc Doan, Souhwan Jung, Daeseon Choi

*These authors contributed equally and are listed in alphabetical order.

RoCo perturbation-code framework
ICASSP 2026OralMay. 2026

RoCo: Robust Code for Fast and Effective Proactive Defense against Voice Cloning Attack

Seungmin Kim, Dain Kim, Sohee Park, Daeseon Choi

RoVo embedding-level voice protection framework
IEEE AccessUnder review2026

RoVo: Robust Voice Protection Against Voice Cloning Attacks via Embedding-Level Adversarial Perturbations

Seungmin Kim, Sohee Park, Dain Kim, Donghyun Kim, Jisu Lee, Daeseon Choi

Session replication attack overview showing the attacker, victim, client, and server
IFIP SEC 2024Jun. 2024

Session Replication Attack Through QR Code Sniffing in Passkey CTAP Registration

Donghyun Kim, Seungmin Kim, Gwonsang Ryu, Daeseon Choi

Explore the research behind the papers

Research projects

2026 – 2028

Real-time Deepfake Disruption via Adversarial Noise Injection

Personal Information Protection Commission (PIPC)

Proactive defense against voice-cloning attacks. Led proposal preparation.

2024 – 2027

Countermeasure Technologies for Generative-AI Security Threats

IITP

Deepfake defense technologies. Participated in proposal preparation.

2021 – 2026

Robust AI & Distributed Attack Detection for Edge AI Security

IITP

DeepVoice detection and voice-deepfake security for smart speakers; built a prototype edge-AI attack-simulation system.

More about my work

Full publication list, projects, and academic background.

Download CV (PDF)