Michael Duan

I am a PhD student at the University of Southern California, advised by Xiang Ren.

My current research interests are AI safety and privacy, tracking data provenance in complex modern AI systems, and interesting ways to make data more accessible (e.g., non-parametric or collaborative learning)

Previously, I graduated with a BS in CS from the University of Washington, where I was fortunate to begin my research in NLP with the fantastic Sewon Min and Niloofar Mireshghallah.

Before that, I had a wonderful start to research with Jon Froehlich at the Makeability Lab, where I worked on vision-based techniques to identify sidewalk accessibility features/challenges.

Email / GitHub / Google Scholar / LinkedIn

Research

	Privasis: Synthesizing the Largest 'Public' Private Dataset from Scratch Hyunwoo Kim, Niloofar Mireshghallah, Michael Duan, Rui Xin, Shuyue Stella Li, Jaehun Jung, David Acuna, Qi Pang, Hanshen Xiao, G. Edward Suh, Sewoong Oh, Yulia Tsvetkov, Pang Wei Koh, Yejin Choi arXiv preprint, 2025
	Frustratingly Simple Retrieval Improves Challenging, Reasoning-Intensive Benchmarks Michael Duan, Xinxi Lyu, Rulin Shao, Pang Wei Koh, Sewon Min arXiv preprint, 2025 arxiv / code /
	A False Sense of Privacy: Evaluating Textual Data Sanitization Beyond Surface-Level Privacy Leakage Rui Xin, Niloofar Mireshghallah, Shuyue Stella Li, Michael Duan, Hyunwoo Kim, Yejin Choi, Yulia Tsvetkov, Sewoong Oh, and Pang Wei Koh NeurIPS Safe Generative AI Workshop, 2025 arxiv / code /
	Do Membership Inference Attacks Work on Large Language Models? Michael Duan, Anshuman Suri, Niloofar Mireshghallah, Sewon Min, Weijia Shi, Luke Zettlemoyer, Yulia Tsvetkov, Yejin Choi, David Evans, Hannaneh Hajishirzi COLM 2024, 2024 arxiv / code /
	Scaling Crowd+AI Sidewalk Accessibility Assessments: Initial Experiments Examining Label Quality and Cross-city Training on Performance Michael Duan, Shosuke Kiami, Logan Milandin, Johnson Kuang, Michael Saugstad, Maryam Hosseini, Jon E. Froehlich ASSETS 2022, 2022 arxiv / code /
	Sidewalk Gallery: An Interactive, Filterable Image Gallery of Over 500,000 Sidewalk Accessibility Problems Michael Duan, Aroosh Kumar, Michael Saugstad, Aileen Zeng, Ilia Savin, Jon E. Froehlich ASSETS 2021, 2021 arxiv /

Design and source code from Jon Barron's website

Michael Duan

Research

Privasis: Synthesizing the Largest 'Public' Private Dataset from Scratch

Frustratingly Simple Retrieval Improves Challenging, Reasoning-Intensive Benchmarks

A False Sense of Privacy: Evaluating Textual Data Sanitization Beyond Surface-Level Privacy Leakage

Do Membership Inference Attacks Work on Large Language Models?

Scaling Crowd+AI Sidewalk Accessibility Assessments: Initial Experiments Examining Label Quality and Cross-city Training on Performance

Sidewalk Gallery: An Interactive, Filterable Image Gallery of Over 500,000 Sidewalk Accessibility Problems