AAAI'26 WORKSHOP

LLM-based Multi-Agent Systems: Towards Responsible, Reliable, and Scalable Agentic Systems (LaMAS 2026)

27 January, 2026, Singapore EXPO

LaMAS 2026 focuses on the emerging field of multi-agent systems powered by LLMs. As interest in using multiple LLM agents to solve complex problems grows, our main objective is to systematically address the critical challenges and opportunities that arise from their interaction. We aim to bridge the gap in understanding failure modes, alignment challenges, and responsible behavior in these systems. The workshop will foster discussion on making these systems powerful, transparent, verifiable, and aligned with human intent.

Call for Paper

contact us at: aaai-lamas2026@googlegroups.com

We welcome both short papers (up to 4 pages) and long papers (up to 8 pages) following the AAAI template. Submissions may include recently published work, under-review papers, work in progress, and position papers. All submissions will undergo peer review through a double-blind process. While workshop publication is non-archival, accepted papers will be featured on our website with the author's permission. Topics of interest include but are not limited to the following:

Organization of multiple LLM agents, including their interaction paradigms, coordination strategies, and communication protocols.
Evaluation and Optimization for assessing LaMAS's performance with emergent behaviors, and fine-tuning techniques to enhance its efficiency and effectiveness.
Safety, Trust, and Responsibility about how to design LaMAS that are safe, aligned, and trustworthy. This includes responsible agent behavior, human-in-the-loop oversight, and regulatory frameworks that ensure accountability, fairness, and transparency in real-world deployments.
Real-World Applications spanning software development, scientific research, education, and business processes, with infrastructures for large-scale multi-agent LLM deployments.

Submission site: https://openreview.net/group?id=AAAI.org/2026/Workshop/LaMAS

Important Dates

Paper Submission Start: 25 Sep 2025 (AoE)

Abstract Registration Due: 27 Oct 2025 (AoE)

Paper Submission Due: 3 Nov 2025 (AoE)

Notification of Paper Acceptance: 14 Nov 2025 (AoE)

Updated: Notification of Paper Acceptance: 21 Nov 2025 (AoE)

Camera-Ready Paper Due: 6 Jan 2026 (AoE)

Workshops Date: 27 Jan 2026, 9 AM - 6 PM

Detailed Workshop Schedule

The detailed schedule is continuously updating.

9:00-9:05

Weinan Zhang

Professor @ Shanghai Jiao Tong University

Opening

9:05-9:45

Bruce Yang

Co-Founder & CEO @ Agnes AI, Singapore

Keynote:

9:45-10:00

Botao Amber Hu

University of Oxford

Some[body] Needs to Receive That Pain: A Position on Agentic Accountability.

We're entering a paradigm shift from autonomous agents to sovereign agents—AI systems that can hold private keys, manage assets, fork themselves, and persist beyond any platform's control. If you encounter agents in the future "agentic web," who's in front of you? Can you trust that agent?

It's starting to look less like software and more like digital wildlife: emergent, unpredictable, ungovernable, and hiding a shoggoth behind a friendly mask. But here's the hard question: Who is accountable when an ungovernable sovereign agent causes harm?

LLMs are playing the Imitation Game—they are not mortal beings. They cannot truly feel pain, fear death, or carry consequences forward. After training, they are static files. Slashing an agent or downgrading its reputation doesn't create any real "urge to change" in their neural system, because nothing inside the model can suffer—just external fakeable memory as their fact. If agents cannot feel pain, some [body] must, responsibly—otherwise, eventually somebody will, unexpectedly.

10:00-10:15

TBD

Oral Pre 1：WebArbiter: A Generative Reasoning Process Reward Model for Web Agents

10:15-10:30

TBD

Oral Pre 2：SAGE: A Theory-Informed LLM-Based Multi-Agent Recommendation System for Grounded AI Solutions Across Domains

10:30-11:00

Coffee Break & Poster Session 1

11:00-11:40

Michael Wooldridge

Ashall Professor @ University of Oxford

Keynote:

11:40-11:55

TBD

Oral Pre 3：Proactive Interference Reveals Working Memory Limits in LLMs Beyond Context Length

11:55-12:10

TBD

Oral Pre 4：The Geometry of Dialogue: Graphing Language Models to Reveal Synergistic Teams for Multi-Agent Collaboration

12:10-12:25

TBD

Oral Pre 5：Federated Agent Reinforcement Learning

12:25-14:00

Lunch

14:00-14:40

Elham Tabassi

Director @ the Brookings Artificial Intelligence and Emerging Technology (AIET) Initiative

Senior Research Scientist @ the National Institute of Standards and Technology (NIST)

Keynote:

14:40-14:55

TBD

Oral Pre 6: Yes FLoReNce, I Will Do Better Next Time! Agentic Feedback Reasoning for Humorous Meme Detection

14:55-15:10

TBD

Oral Pre 7: Disagreements in Reasoning: How a Model's Thinking Process Dictates Persuasion in Multi-Agent Systems

15:10-15:25

TBD

Oral Pre 8: PALADIN: Self-Correcting Language Model Agents to Cure Tool-Failure Cases

15:25-16:00

Coffee Break & Poster Session 2

16:00-16:15

TBD

Oral Pre 9: A Multi-Agent Framework for Enterprise Tool Creation

16:15-16:30

TBD

Oral Pre 10: DYNO : Dynamic Neurosymbolic Orchestrator for Multi-Agent Systems

16:30-17:15

Gopal Ramchurn @ University of Southampton (Host)

Wan Sie Lee @ IMDA, Singapore

Stefano V. Albrecht @ DeepFlow London

Ramayya Krishnan @ Carnegie Mellon University

Mengyue Yang @ University of Bristol

Panel Discussion

We will open the floor to have a discussion about fundamental questions, future directions, and industrial applications of LLM-based multi-agent systems.