Three New Research Papers Explore Multi-Agent AI Systems and Data Management

Three New Research Papers Explore Multi-Agent AI Systems and Data Management

Three recent preprints on arXiv address different aspects of multi-agent AI systems and their applications.

According to arXiv:2602.21351v1, researchers have developed a hierarchical multi-agent system designed for autonomous discovery in geoscientific data archives. The paper addresses what it describes as a “significant scalability challenge” created by the rapid accumulation of Earth science data. The authors note that while repositories like PANGAEA host vast collections of datasets, “citation metrics indicate that a substantial portion remains underutilized, limiting data” reuse.

A separate paper (arXiv:2602.13477v2) titled “OMNI-LEAK” examines security concerns in multi-agent systems. The research focuses on “Orchestrator Multi-Agent Network Induced Data Leakage.” According to the abstract, while prior work has examined safety and misuse risks associated with individual LLM agents, this paper addresses risks specifically related to their coordinated use in multi-agent systems.

Finally, arXiv:2602.21255v1 presents theoretical work establishing “a general equilibrium theory for systems of large language model (LLM) agents operating under centralized orchestration.” The framework extends the Arrow-Debreu production economy model from 1954 “to infinite-dimensional commodities,” according to the paper.

All three papers represent new submissions to arXiv’s AI section.