Naixin Zhai 翟乃鑫

nicyzhai [at] gmail.com

I am a final-year M.Eng. candidate at the University of Science and Technology of China (USTC), under the guidance of Xun Yang and Pengyang Shao. Before that, I received a B.Eng. Degree from the Hefei Univerisity of Technology (HFUT).

My research interests primarily lie in LLM Unlearning and Safety on AI-Generated Content.

Google Scholar | Github | ORCID | LinkedIn | WeChat | Album

News

[Jul 2026] PALU is awarded as the Outstanding Paper (Top 0.2%)!
[Apr 2026] Our paper PALU is accepted by ACL 2026 and selected as an Oral Presentation!

Publications

[NEW] Maximizing Local Entropy Where It Matters: Prefix-Aware Localized LLM Unlearning
Naixin Zhai, Pengyang Shao, Binbin Zheng, Yonghui Yang, Fei Shen, Long Bai, Xun Yang
[paper] [code] 🏆 ACL 2026 Outstanding Paper Award

BalDRO: A Distributionally Robust Optimization based Framework for Large Language Model Unlearning
Pengyang Shao, Naixin Zhai, Lei Chen, Yonghui Yang, Fengbin Zhu, Xun Yang, Meng Wang
[paper] [code] WWW 2026 Web4Good Oral Presentation

Vulnerabilities in AI-Generated Image Detection: The Challenge of Adversarial Attacks
Yunfeng Diao*, Naixin Zhai*, Changtao Miao, Zitong Yu, Xingxing Wei, Xun Yang, Meng Wang
[paper] [code] TMM 2026

Projects

Awesome AI-generated Image Detection
A curated collection of datasets, tools, papers, and code for AI-generated image detection.
[Repository] 104 stars

Talks

[NEW] Jul 2026: Invited talk on PALU (ACL26) at BAAI Hub, NICE AI Talk

Awards

University-level Excellent Student Award, HFUT, 2023
Wang Zhenghu Entrepreneurship Scholarship, HFUT, 2023
National Scholarship, HFUT, 2022