I am currently a postdoctoral researcher at CISPA Helmholtz Center for Information Security. Before that, I earned my Ph.D. at CISPA in 2025, co-supervised by Prof. Michael Backes and Prof. Yang Zhang. I obtained my master’s degree from Shanghai Jiao Tong University in 2021 and my bachelor’s degree from Shandong University in 2018.

My research interests lie in trustworthy machine learning, with a focus on the safety and security risks of generative AI. I work on understanding these risks, both unintentional and malicious, particularly in large generative models, and I actively develop mitigations to defend against the misuse of generative AI and the spread of harmful content across social media. Broadly, my recent work can be categorized into three main directions.

• Understanding the Real-World Misuse of Generative AI, such as AI-Generated unsafe imagery, optical illusions, and hate speech.
• Identifying and Mitigating Novel Security Risks of Generative AI, including implicit jailbreaking and prompt stealing attacks.
• Actively Mitigating the Risks of Harmful Media and Deepfakes, such as tracing online hateful memes, detecting fake images and fake point clouds to enhance content safety and media integrity.

What’s New

• [2025.06-07] Three first-authored papers got accepted in USENIX 2025, ICCV 2025, and CCS 2025.
• [2025.05] I successfully passed my PhD defense and became Dr. Qu.
• [2025.01] Two papers got accepted in USENIX 2025. See you in Seattle!
• [2024.02] Our paper titled “Prompt Stealing Attacks Against Text-to-Image Generation Models” got accepted in USENIX 2024. See you in Philadelphia!
• [2024.01] One paper titled “FAKEPCD: Fake Point Cloud Detection via Source Attribution” got accepted in ACM ASIACCS 2024. See you in Singapore!
• [2023.05] One paper titled “Unsafe Diffusion: On the Generation of Unsafe Images and Hateful Memes From Text-To-Image Models” got accepted in ACM CCS 2023! See you in Copenhagen!
• [2023.01] Our paper titled “Prompt Stealing Attacks Against Text-to-Image Generation Models” is online. This is the first study on prompt stealing attacks. You can read it here!
• [2022.12] One paper titled “On the Evolution of (Hateful) Memes by Means of Multimodal Contrastive Learning” got accepted in Oakland 2023!
• [2022.11] I passed my PhD Qualifying Exam!
• [2021.11] I joined CISPA to start my PhD!