Embedded Files
About
LAW 2025 invites high-quality submissions that explore the intersection of Large Language Models (LMs), Agent Models (AMs), and World Models (WMs). The workshop seeks innovative research aimed at integrating these areas to develop AI systems capable of sophisticated reasoning, planning, simulation, and effective interaction within dynamic environments.
We accept non-archival submissions, which will be hosted on our workshop website and OpenReview. Papers accepted to our workshops will be non-archival as well.
All papers must be submitted in NeurIPS format via the OpenReview submission portal. Submissions should be anonymized according to NeurIPS guidelines. Authors may freely share their work as preprints before or during review. Papers concurrently submitted to other conferences (e.g., ICLR) are explicitly encouraged.
Sponsored by Lambda
🏆 Best Paper Prize: We are pleased to announce that the workshop will feature the Best Paper Prize (One best paper, three Honorable Mentions) to recognize outstanding research contributions, generously sponsored by Lambda.
Beyond the public recognition and spotlight presentations at the workshop, every accepted paper will receive compute credits, with award-winning papers earning a larger allocation to supercharge their next research cycle. All submissions are automatically considered, reflecting our commitment to advancing impactful work at the intersection of language, agents, and world models.
Prize papers will be selected based on their excellence, originality, and impact. We encourage all participants to submit their best work for consideration!
Important Dates (Anywhere on Earth, AoE)
Submission Opens: July 15, 2025 - OpenReview Submission Link
Submission Deadline: September 1, 2025 [EXTENDED DEADLINE!]
Round 2 Submission Deadline: September 20, 2025 - OpenReview Submission Link
Notification of Acceptance: September 22, 2025
Camera-Ready Deadline: October 31, 2025
Workshop Date: December 7, 2025
Topics of Interest
We welcome submissions addressing theoretical, methodological, and empirical research across these themes, including but not limited to:
Internal world representations within LLMs, and approaches to probe or enhance them.
Novel architectures combining language models with symbolic, probabilistic, multimodal, or simulation-based world models.
LLM-driven reinforcement learning and planning agents: capabilities, limitations, and innovations.
Benchmarks, evaluation methodologies, and diagnostic tasks bridging language understanding, action generation, and simulation.
Safety, alignment, and socio-economic implications of agentic systems integrating world models.
Theoretical frameworks uniting cognition, language grounding, and control.
Scalable simulation platforms and synthetic data generation for joint research in LLMs, agents, and world modeling.
Real-world applications in robotics, scientific discovery, human-AI interaction, gaming, and other open-ended tasks.
Submission Tracks
Submissions are invited in the following categories:
Research Papers: Present original algorithms, analyses, and empirical studies (4–9 pages).
Benchmark Papers: Introduce new datasets, evaluation frameworks, or environments (4–9 pages).
Position/Review Papers: Offer visionary perspectives or comprehensive surveys (4–9 pages).
Demo Papers: Showcase practical demonstrations for practical agent and world models (4–9 pages).
Submission Instructions
Submission Instructions
We kindly note that vision conferences such as CVPR and ICCV consider peer-reviewed workshop papers as publications if their length exceeds 4 pages (excluding references), even if they do not appear in a proceedings. If you are considering submitting your paper later to vision conferences, please consider submitting an abridged version as a 4-page short paper.
Review Process:
Double-blind review via OpenReview.
Papers reviewed based on clarity, originality, methodological rigor, and explicit connection among language, agent, and world models.
Desk rejection policy for papers that significantly overlap with previously published archival works.
Presentation and Awards
Spotlight presentations (15 minutes) for top submissions.
Poster sessions for all accepted papers.
Optional pre-recorded videos to maximize accessibility.
Contact: law2025@googlegroups.com
Page updated
Google Sites
Report abuse