Ph.D. Candidate · Trustworthy AIGC
Kun Xu 许锟
I study generative model security, uncertainty, and trustworthy AI, with a focus on concept-level risks in text-to-image generation and deepfake detection.
- NUAAPh.D. candidate in Cyberspace Security
- NanjingBased at NUAA in Nanjing, China
- FocusGenerative concept security
About
A research profile built around safer generative AI.
I am a Ph.D. candidate in Cyberspace Security at Nanjing University of Aeronautics and Astronautics (NUAA), advised by Prof. Yushu Zhang. I have successfully completed my Ph.D. dissertation defense and expect to receive my Ph.D. in October 2026. I am currently continuing my research at NUAA in Nanjing, China.
From September 2025 to August 2026, I conducted visiting doctoral research in Milan, Italy, supported by the China Scholarship Council (CSC) Excellence Talent Programme and supervised by Prof. Elena Ferrari. During the visit, I also engaged in academic exchange and research collaboration with Prof. Pierangela Samarati and Prof. Vincenzo Piuri at the University of Milan, further broadening my international research experience in trustworthy AI, security, and privacy.
My research focuses on the safety, evaluation, and reliability of generative AI, including concept-level risk assessment, malicious and toxic content detection, uncertainty and calibration in diffusion models, privacy-aware synthesis, and deepfake forensics. My work has appeared in or been accepted by venues including IEEE TDSC and ACM Multimedia, and I am currently seeking postdoctoral or faculty opportunities.
Research
Research themes
A compact map of the topics that connect my current work.
News
Recent updates
Selected milestones, submissions, talks, and awards.
Publications
Publications and selected manuscripts
Filter by year, search by keyword, and copy BibTeX for each listed work.
Projects
Research highlights
A more visual view of methods, artifacts, and active research directions.
CV
Academic timeline
A concise overview of my education, honors, academic service, and professional activities.
Education
Honors
Academic Service
Contact
Open to research conversations and collaboration.
I welcome conversations about trustworthy AIGC, generative model security, concept-level safety, uncertainty and calibration, and multimedia forensics.