About the Project
A fully funded PhD opportunity to participate in the world-leading research undertaken by the EPSRC Doctoral Landscape Award at the University of Sheffield. We aim to develop multi-modal Large Language Models (LLMs) for AI-Generated Content Detection and Watermarking.
Supervised by Dr Delvin Ce Zhang and Dr Zhixiang Chen.
The University of Sheffield is one of the leading Russell Group universities in the UK. We carry out cutting-edge research with strong links to industry. When you enrol to do a PhD with us, you will be working with world-leading academics and have access to top of the range facilities. As a PhD student you will have the opportunity to gain skills not only to conduct research, but also to take your career to the next level, whether you want to stay in academia, go into industry or the public sector, or set up your own company. You will have access to a range of training and support services to help you excel in your studies and beyond.
Project Description
Social media platforms such as Twitter/X, Facebook, Instagram, and TikTok have become major channels for people to access and share information. However, information circulated on social media is not always reliable. With the rapid development of Generative AI, users are increasingly exposed to AI-Generated Content, including ChatGPT-generated texts, synthetic images, and videos produced by models such as Sora. Although AI-generated content has many legitimate applications, it can also be misused to produce misleading or deceptive information and may contribute to the spread of misinformation online.
For ordinary social media users, it can be difficult to determine whether a piece of content has been created by a human or generated by an AI system. Most users do not have sufficient technical expertise to identify subtle linguistic, visual, or multimodal characteristics that may reveal the origin of online content. Furthermore, as generative AI models continue to improve, AI-generated content is becoming increasingly realistic and more difficult to distinguish from human-produced content. Misleading synthetic content may therefore create unnecessary concerns, influence public understanding, or reduce trust in online information.
Motivated by this real-world challenge, this project investigates AI-Generated Content Detection and Watermarking, focusing on two closely related research directions.
The first direction concerns automatic AI-generated Content Detection. We aim to develop a Vision-Language Model that can analyze social media content and determine whether it is more likely to be human-produced or AI-generated. The model will investigate textual, visual, and multimodal information, including linguistic patterns, visual characteristics, semantic consistency, and relationships between different modalities. When potentially AI-generated content is detected, the system could provide an appropriate indication or warning to users, encouraging them to interpret the information more cautiously.
The second direction focuses on Watermarking for Generative AI. We plan to investigate the internal mechanisms and generation processes of generative AI models and design watermarking techniques that embed identifiable signals into AI-generated content. These watermarks should ideally be invisible or imperceptible to human users while remaining detectable by computational methods. They could therefore serve as reliable machine-readable indicators of AI-generated provenance. The project will also investigate the robustness of such watermarks under common transformations and modifications.
By combining detection and watermarking, this project aims to provide complementary mechanisms for identifying AI-generated content. Detection enables the analysis of content with unknown origins, while watermarking introduces provenance information during generation. Together, these directions seek to contribute to trustworthy generative AI and a more reliable online information environment.
Supervisors' Bio
Dr Delvin Ce Zhang is a Lecturer (Assistant Professor) at University of Sheffield. His research centers around applying Multi-Modal and Retrieval-Augmented Large Language Models for many real-world problems, including Misinformation Detection, Information Retrieval, and AI for Science. Currently, he serves as the Poster Chair for KDD-27 conference. Earlier, he served as the Technical Chair for DSAA-25 conference. His papers are mainly published on NeurIPS, ICML, AAAI, KDD, ACL, EMNLP, NAACL, TKDE, etc. He was a Postdoctoral Researcher at The Pennsylvania State University and Yale University. He received PhD degree at Singapore Management University. More information can be found at: www.delvincezhang.com.
Dr Zhixiang Chen is a Lecturer (Assistant Professor) at University of Sheffield. Previously, he was a Postdoctoral Research Associate at Imperial Computer Vision and Learning Lab of Imperial College London working with Dr Tae-Kyun Kim and a Postdoctoral researcher at Department of Automation of Tsinghua University working with Prof Jie Zhou and Prof Jiwen Lu. He received his PhD from Tsinghua University under the supervision of Prof Jie Zhou and Prof Jiwen Lu. His research interest lies in 3D vision on human - the perception of 3D humans by machine and large scale retrieval. His papers are published on NeurIPS, AAAI, CVPR, ICCV, ECCV, WACV, etc.
How to apply
Interested candidates are strongly encouraged to contact the project supervisors (Dr Delvin Ce Zhang, [contact details available in the full listing]), including [PhD Application] in the email subject line, to discuss your interest in and suitability for the project prior to submitting your application.
Please refer to the EPSRC DLA webpage for detailed information about the EPSRC DLA and how to apply.
Applicants must ensure that they use the correct application form linked to the EPSRC DLA webpage and must submit an application for October 2027 entry. The form is unique to the EPSRC DLA. It contains the list of advertised projects from which to select, and three questions specific to the DLA.
The application deadline is 17:00 GMT 13 January 2027.
The majority of our DLA studentships will be allocated to candidates who qualify for Home fee status, in accordance with EPSRC funding regulations which place strict limits on international recruitment.
Funding Notes
The award will fund the full tuition fee and a maintenance stipend at the UKRI rate (currently £21,805 per annum) for 3.5 years, as well as a research grant to support costs associated with the project.
