Publications

Xiangyu Zhou, Saleh Zare Zade, Rafi Ibn Sultan, Alexander Kotov, Dongxiao Zhu (2026). Towards Mitigating Deceptive Safety Alignment in Large Reasoning Models. NeurIPS-26.

Source Document Code DOI

Equal contribution

Equal contribution

, Saleh Zare Zade, Prashant Khanduri, Dongxiao Zhu (2026). Hijacking Large Language Models via Adversarial In-Context Learning. ECML-PKDD-26.

Source Document Code Preprint

Xiangyu Zhou, Yao Qiang, Saleh Zare Zade, Douglas Zytko, Prashant Khanduri, Dongxiao Zhu (2026). Not all tokens are meant to be forgotten. AAAI-26.

Source Document Code DOI

Rafi Ibn Sultan, Hui Zhu, Xiangyu Zhou, Chengyin Li, Prashant Khanduri, Marco Brocanelli, Dongxiao Zhu (2026). WalkGPT Grounded Vision-Language Conversation with Depth-Aware Segmentation for Pedestrian Navigation. CVPR-26.

Source Document Code Dataset

Saleh Zare Zade, Xiangyu Zhou, Sijia Liu, Dongxiao Zhu (2026). Attention Smoothing Is All You Need For Unlearning. ICLR-26.

Source Document Code

Emma Walquist, Isha Datey, Wenqi Zheng, Xiangyu Zhou, Kelly Berishaj, Melissa Mcdonald, Michele Parkhill, Dongxiao Zhu, Douglas Zytko (2025). Collective Consent Who Needs to Consent to the Donation of Data Representing Multiple People?. CSCW-25.

Source Document DOI

Saleh Zare Zade, Yao Qiang, Xiangyu Zhou, Hui Zhu, Mohammad Amin Roshani, Prashant Khanduri, Dongxiao Zhu (2025). Automatic calibration for membership inference attack on large language models. ECAI-25.

Source Document

Mohammad Amin Roshani, Xiangyu Zhou, Yao Qiang, Srinivasan Suresh, Steven Hicks, Usha Sethuraman, Dongxiao Zhu (2025). Generative large language model—powered conversational ai app for personalized risk assessment Case study in covid-19. JMIR.

Source Document DOI

Wenqi Zheng, Emma Walquist, Isha Datey, Xiangyu Zhou, Kelly Berishaj, Melissa Mcdonald, Michele Parkhill, Dongxiao Zhu, Douglas Zytko (2024). “It’s not what we were trying to get at, but I think maybe it should be” learning how to do trauma-informed design with a data donation platform for online dating sexual violence. CHI-24.

Source Document DOI

Equal contribution

Equal contribution

, Saleh Zare Zade, Mohammad Amin Roshani, Prashant Khanduri, Douglas Zytko, Dongxiao Zhu (2024). Learning to poison large language models for downstream manipulation.