I am currently a Proleptic Lecturer in AI Safety at Uniersity of York. My research focuses on Large Language Model Safety and AI Security, particularly around security risks in modern AI systems, e.g., Backdoor Attacks, Membership Inference Attacks, Model Extraction Attacks in Large Language Models, Graph Neural Networks, and Multi-Agent Systems. I’ve mostly worked at the intersection of machine learning and security, with the purpose of making modern AI systems more Robust, Secure, and Private.

Before that, I was a Postdoc researcher with the SprintML Lab at CISPA, working with Dr. Franziska Boenisch and Dr. Adam Dziedzic. Prior to that, I obtained my PhD degree with the Cybersecurity group at Delft University of Technology under the supervision of Prof.dr.ir. Inald Lagendijk, Dr. Stjepan Picek and Dr. Frans A. Oliehoek.

I am looking for motivated PhD students to join my team working on (i) AI security; (ii) multi-agent systems; (iii) graph machine learning. Fully funded PhD positions are available through:

• UKRI AI Centre for Doctoral Training in Safe Artificial Intelligence Systems (SAINTS)
• China Scholarships Council - University of York PhD Scholarship

If you are interested in applying, please send me an email with:

• CV
• Transcript with publication list (if available)
• A brief description of your research interests

Research Interests

• AI Security (LLM safety/security, adversarial attacks/defense)
• Graph Machine Learning (graph neural networks, knowledge graph)
• Multi-Agent Systems

What’s New

• [2026.04] I joined the University of York as a Proleptic Lecturer in AI Safety!
• [2026.04] Joined the Program Committee of ESORICS’26.
• [2026.03] Our paper titled “ADAGE: Active Defenses Against GNN Extraction” was accepted by AsiaCCS 2026!
• [2026.02] I gave a talk in the Natural Language Processing (NLP) group, Aalborg University, Copenhagen, Denmark.
• [2026.01] Our paper titled “Frequency-Domain Model Fingerprinting for Image Autoregressive Models” was accepted by AAAI 2026 Workshop AIGOV!
• [2025.09] Our paper titled “Finding Memo(rization) in Graph Neural Networks” was accepted by NeurIPS 2025!
• [2025.05] Our paper titled “Efficient and Privacy-Preserving Soft Prompt Transfer for LLMs” was accepted by ICML 2025!

Selected Publications

• ADAGE: Active Defenses Against GNN Extraction. Jing Xu, Franziska Boenisch, Adam Dziedzic; AsiaCCS 2026. [arXiv]
• Memorization in Graph Neural Networks. Adarsh Jamadandi*, Jing Xu, Adam Dziedzic, Franziska Boenisch; NeurIPS 2025. [arXiv]
• Efficient and Privacy-Preserving Soft Prompt Transfer for LLMs. Xun Wang*, Jing Xu, Franziska Boenisch, Michael Backes, Christopher A. Choquette-Choo, Adam Dziedzic; ICML 2025. [PDF]
• DP-GPL: Differentially Private Graph Prompt Learning. Jing Xu, Franziska Boenisch, Iyiola Emmanuel Olatunji, Adam Dziedzic; ICLR Workshop on Foundation Models in the Wild 2025. [PDF]
• Watermarking Graph Neural Networks based on Backdoor Attacks. Jing Xu, Stefanos Koffas, Oguzhan Ersoy, Stjepan Picek; Euro S&P 2023. [PDF]
• More is Better (Mostly): On the Backdoor Attacks in Federated Graph Neural Networks. Jing Xu, Rui Wang, Kaitai Liang, Stjepan Picek; ACSAC 2022. [PDF]
• Explainability-based backdoor attacks against graph neural networks. Jing Xu, Minhui(Jason) Xue, Stjepan Picek; WiseML 2021. [PDF]