Publications

Safety in Batches? Understanding and Mitigating Safety Failures in Batch Prompting

Kihyun Kim, Hee-Seon Kim, Wonjun Lee, Changick Kim

Asia-Pacific Chapter of ACL & International Joint Conference on NLP (AACL-IJCNLP Main), 2026

Paper

AIM: Anchor Identity Features, then Match for Multimodal Large Language Model Unlearning

Wonjun Lee*, Jaehyuk Jang*, Kangwook Ko*, Hee-Seon Kim, Changick Kim (* indicates equal contribution)

The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP findings), 2026

Paper | Code

Where Identity Lives: Localized, Retain-Free Identity Unlearning in Multimodal Large Language Models

Kangwook Ko*, Jaehyuk Jang*, Wonjun Lee*, Hee-Seon Kim, Changick Kim (* indicates equal contribution)

The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP findings), 2026

Paper

SubT: Subspace Tuning for Few-shot Generalization of Audio-Language Models

Jaehyuk Jang, Kangwook Ko, Wonjun Lee, Changick Kim

The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP findings), 2026

Paper | Code

Jailbreak to Protect: Buffering Harmful Fine-Tuning via Temporary Jailbreaking LoRA in Large Language Models Spotlight (Top 2.2%)

Seokil Ham, Jaehyuk Jang, Wonjun Lee, Changick Kim

International Conference on Machine Learning (ICML), 2026

Paper | Code | Slides | Poster

Generalizable Prompt Tuning for Audio-Language Models via Semantic Expansion Poster

Jaehyuk Jang*, Wonjun Lee*, Kangwook Ko*, Changick Kim (* indicates equal contribution)

The 64th Annual Meeting of the Association for Computational Linguistics (ACL findings), 2026

Paper | Slides | Poster

Efficient Test-Time Optimization for Depth Completion via Low-Rank Decoder Adaptation

Minseok Seo*, Wonjun Lee*, Jaehyuk Jang, Changick Kim (* indicates equal contribution)

arXiv preprint, 2026

Paper | Code | Project

SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection

Younghun Kim, Minsuk Jang, Myung-Joon Kwon, Wonjun Lee, Changick Kim

arXiv preprint, 2025

Paper

Benign-to-Toxic Jailbreaking: Inducing Harmful Responses from Harmless Prompts

Hee-Seon Kim, Minbeom Kim, Wonjun Lee, Kihyun Kim, Changick Kim

arXiv preprint, 2025

Optimization-based jailbreaking that induces safety misalignment from benign conditioning prompts.

Paper

Safire: Segment Any Forged Image Region Poster

Myung-Joon Kwon*, Wonjun Lee*, Seung-Hun Nam, Minji Son, Changick Kim (* indicates equal contribution)

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2025

Paper | Code | Poster

Friday: Mitigating unintentional facial identity in deepfake detectors guided by facial recognizers Oral

Younghun Kim, Myung-Joon Kwon, Wonjun Lee, Changick Kim

IEEE International Conference on Visual Communications and Image Processing (VCIP), 2024

Paper | Slides