Chanhui Lee

Ph.D. Student

Chanhui Lee

AI safety, adversarial attacks, and generative models

Pohang University of Science and Technology (POSTECH)

01

About

I am Chanhui Lee, a Ph.D. student at POSTECH CVLab, advised by Prof. Jeany Son. My research interests include AI safety, adversarial attacks, and generative models.

Pohang, Korea AI safety Adversarial attack Generative models chanlee@postech.ac.kr
02

Education

Ph.D. student, Pohang University of Science and Technology (POSTECH), advised by Prof. Jeany Son.

M.S., Gwangju Institute of Science and Technology, advised by Prof. Jeany Son.

B.S., Gwangju Institute of Science and Technology, including military service.

03

Publications

ECCV 2026

Universal Image Immunization against Diffusion-based Image Editing via Semantic Injection

Chanhui Lee, Donggyu Choi, Seunghyun Shin, Hae-gon Jeon, Jeany Son

Paper Project Page 
@misc{lee2026universal,
  title={Universal Image Immunization against Diffusion-based Image Editing via Semantic Injection},
  author={Lee, Chanhui and Choi, Donggyu and Shin, Seunghyun and Jeon, Hae-gon and Son, Jeany},
  year={2026},
  eprint={2602.14679},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}

ECCV 2026

Anti-Prompt: Image Protection against Text-Guided Image-to-Video Generation

Yeonghwan Song, Chanhui Lee, Jinsoo Park, Jeany Son

Paper Project Page 
@misc{song2026antiprompt,
  title={Anti-Prompt: Image Protection against Text-Guided Image-to-Video Generation},
  author={Song, Yeonghwan and Lee, Chanhui and Park, Jinsoo and Son, Jeany},
  year={2026},
  eprint={2607.01499},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}

CVPR 2025

Data-free Universal Adversarial Perturbation with Pseudo-semantic Prior

Chanhui Lee, Yeonghwan Song, Jeany Son

Paper Project Page 
@inproceedings{lee2025datafree,
  title={Data-free Universal Adversarial Perturbation with Pseudo-semantic Prior},
  author={Lee, Chanhui and Song, Yeonghwan and Son, Jeany},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2025}
}
04

News

Two papers are accepted to ECCV 2026.

Graduated with an M.S. from GIST AI Graduate School.

One paper is accepted to CVPR 2025.