Harness-Zero: Harness Distillation via Agent-as-Harness
Peking University; Google; The Hong Kong University of Science and Technology
Emergent Collusion in Long-Horizon LLM Agent Interaction
Stanford University; Georgia Tech
Et Tu, Brute? Economic Misalignment in Personal AI Agents
Foundation AI, Cisco; Carnegie Mellon University
BackTrend: Evaluating Scientific Weak-Signal Prediction via Backward Reconstruction
Yale NLP Lab; New York University; TCS Research
A Global Comparison of Schemas, Transparency, and Interoperability in Public-Sector AI Registers and Inventories
Cornell University; University of Toronto
Pinocchio: Fast Uncertainty Estimates for Black-Box Language Models
University of Maryland; Ritual AI; Fudan University; Columbia University
Partner-Specific Affective Precision in Social Active Inference
Mission San Jose High School; University of Chicago
Extracting Arguments, Not Just Classifying Them: Instruction-Tuned LLMs for Generative Component Detection
Université Côte d'Azur; CNRS; INRIA
MedRSI: Recursive Self-Improvement for Medical Agents via Clinically Aligned Self-Evolution
University of Oxford; Stanford University
GRUET: Quantifying Uncertainty of Agentic Reasoning-and-Acting Processes
Nanjing University
Convex AI Compositionality and the Governance of AI System Populations
University of Zürich; SUPSI; ETH Zürich
Construting Reverse Thinking: Developing Large Language Models’ Reverse Thingking Ability
China University of Petroleum (East China); Goertek Inc.; Nankai University
Epi-Logic: A Conceptual Framework for Epistemic Runtime Control, Schema Validity Checking, and Controlled Accommodation in Autonomous AI Agents
Independent Researcher
14.
ワールドステート生成器
World State Generator
University of California, Irvine; Northeastern University
TimeLitmus: A Diagnostic Benchmark for Cross-Modal Understanding and Explanation Faithfulness in Event-Conditioned Time-Series Prediction
Wuhan University; Nanjing Audit University; The University of Manchester; MBZUAI; McGill University; Shanghai Jiao Tong University
Beyond Endpoint Performance: Process-Level Evaluation of Self-Evolving Agents
Zhejiang University; Alibaba Group
DUMA-Bench: A Dual-Control Multi-Agent Benchmark for Evaluating LLM Agent Security
ITMO University; Hive Trace Lab; Independent Researcher
Custom Named Entity Recognition and Topic Classification for Global Health Publications
Faculty of Sciences
Ascent: An Agentic System over the Model Context Protocol for Real-World Clinical Data Analysis
Bayer AG
The Endless Exam: Mathematical Constructions from Today’s Models toward Superintelligence
The University of Texas at Arlington
