Yali Du
Associate Professor in AI · King's College London · Turing Fellow · ELLIS Member
Bush House, 30 Aldwych
London WC2B 4BG
yali.du@kcl.ac.uk
I am a Senior Lecturer (Associate Professor) in AI at King’s College London, where I head the Distributed AI Group and lead the Cooperative AI Lab. I am also a Turing Fellow and secondee at The Alan Turing Institute, and a member of the European Laboratory for Learning and Intelligent Systems (ELLIS).
My research develops cooperative and safe AI agents that can learn, coordinate, and align with humans in complex decision-making environments, with a focus on scalable multi-agent learning, human–AI coordination, and value alignment.
I received the AAAI New Faculty Highlights Award and was named a Rising Star in AI by KAUST in 2023. I serve as Senior Area Chair for AAMAS and NAACL, Area Chair for NeurIPS, ICML, ICLR and IJCAI, and Associate Editor for IEEE Transactions on Artificial Intelligence, and have served on the organising committees of AAMAS 2023, NeurIPS 2024, DAI 2025 and AAMAS 2027.
Research: Safe and Cooperative AI
Three connected directions in the Cooperative AI Lab. The goal: agents that can learn, coordinate and remain safe in multi-agent and human-centred environments.
Evaluating agent societies
What emerges when many agents interact under mixed motives?
- SocialJax (ICLR '26)
- OpenSocialJax (arXiv '26)
- LLM agent populations (AAMAS '25)
Learning & aligning
How can agents coordinate at scale, with partners and with humans?
- Scalable networked MARL (Nature MI '24, NeurIPS '25)
- Zero-shot human–AI coordination (NeurIPS '23)
- Learning from human and language feedback (ICML '25, EMNLP '25)
Safety & oversight
How do we keep agent societies safe and accountable?
- Delegation safety: BazaarBench (arXiv '26)
- Safe MARL with language constraints (AAAI '26)
- Multi-agent AI safety and governance
Full publication list: Cooperative AI Lab publications · Google Scholar
Announcements
- We run the Agentic AI Frontier Online Seminar every Friday at 9 AM PT. Check the upcoming sessions and subscribe to the mailing list.
- We receive support from the UK AI Security Institute for our research on the systemic safety of large language models, with a focus on the emergent behaviours of LLM agent groups. One postdoc position is funded.
- SocialJax, a suite of sequential social dilemma environments for multi-agent RL in JAX, developed with Joel Leibo (Google DeepMind).
News
| Dec 2026 | Organising the ELLIS UnConference on Multi-Agent AI for Scientific Discovery. |
| Nov 2026 | Invited talk at the Department of Chemistry, University of Liverpool. |
| Nov 2026 | Invited keynote at the LLM Agents for Social Simulation Workshop at CIKM 2026. |
| Oct 2026 | Invited speaker at the HKIAS International Conference on Multi-Agent Systems and AI 2026; academic visits to HKUST, PolyU and HKBU. |
| Oct 2026 | Invited talk at the Oxford Control Seminar Series, Control Group, University of Oxford. |
| Jun 2026 | Invited to serve as Senior Area Chair for NAACL 2027 and AAMAS 2027. |
| Mar 2026 | Two Responsible AI UK AI & Robotics Best Research Paper Awards. |
| Jan 2026 | Joined ELLIS as an ELLIS Member. |