Tool use, function calling, orchestration and the protocols that let models act rather than only answer.
arXiv:2609.05513v1 Announce Type: new Abstract: Web agents have achieved significant success in automating complex internet tasks but deploying them in real-world environments requires continuous onl...
arXiv:2511.09057v4 Announce Type: replace-cross Abstract: A world model is a cognitive simulator of the real-world environment allowing biological agents to reason about how the world evolves, whethe...
The paper investigates whether large language model (LLM) agents can simulate public deliberation by reflecting population opinion patterns and producing interaction-driven opinion change. Using census‑grounded Korean personas debating real policy questions, the study finds that persona agents fail to reliably reproduce population opinion patterns, often concentrating responses and reversing demographic differences. While deliberations generate reasoned, reciprocal arguments and some stance movement, much of this change occurs without peer exchange, and anchoring agents to population‑informed starting positions suppresses updating, indicating that population representation, argument generation, and interaction‑driven opinion change do not necessarily align.
arXiv:2605.23220v2 Announce Type: replace Abstract: Despite the growing use of world models as decision-making agents, their adversarial robustness remains underexplored due to the lack of dedicated...
arXiv:2609.05576v1 Announce Type: new Abstract: The paradigm of LLMs has rapidly shifted from passive language interfaces to autonomous Claw-like agents that execute long-horizon tasks across statefu...
arXiv:2609.05587v1 Announce Type: new Abstract: Existing evaluations of tool-using agents primarily measure whether an agent can successfully complete diverse tasks with tools. These evaluations gene...
arXiv:2604.16625v2 Announce Type: replace-cross Abstract: Recent large language model (LLM) agents have shown promise in using execution feedback for test-time adaptation. However, robust self-improv...
arXiv:2609.06783v1 Announce Type: cross Abstract: LLM agents operate in workflows where unsafe actions can have real consequences. Existing safety evaluations often reduce behavior to a single score,...
arXiv:2609.06835v1 Announce Type: cross Abstract: Agentic AI systems execute complex tasks through long-horizon workflows of planning, tool use, and multi-agent coordination. Task failures in these s...
GoAnt is a quality‑diversity multi‑agent search framework designed for discovering alpha factors in market microstructure data. It employs non‑communicating Explorer, Exploiter, and Connector workers that share an adaptive Mental Map and a Queen dispatcher to allocate evaluation budgets efficiently. On real A‑share data from 2023‑2026, GoAnt achieves significantly higher quality‑weighted yields than baseline methods and maintains strong out‑of‑sample performance.
ClosureBound is a reference monitor that enforces authorization boundaries for agent skills by binding each grant to an exact dependency closure, effect ceiling, purpose, validity, and epochs. It resolves typed graph nodes, normalizes operations into an external‑effect IR, and admits actions only when a joint witness satisfies all bounds, ensuring metadata non‑authority, closure determinism, and other security properties. Empirical evaluation on 549 public skills shows many lack proper dependency declarations, underscoring the need for conservative closure discovery and broader runtime validation.
The paper reports the first Turing test for speech‑to‑speech systems, gathering 2,968 human judgments on conversations between nine state‑of‑the‑art S2S systems and 28 humans. None of the evaluated systems passed the test, highlighting a clear gap in human‑likeness. The authors diagnose the failure with an 18‑dimension taxonomy, finding that paralinguistic cues, emotional expressivity, and conversational persona—not semantic understanding—are the main bottlenecks, and they propose an interpretable model for automatic human‑vs‑machine discrimination.
The paper proposes a CSI‑free hierarchical multi‑agent reinforcement learning framework for controlling reconfigurable reflective surfaces in millimeter‑wave networks. By replacing per‑element channel estimation with user localization data, the system uses a two‑tier neural architecture: a high‑level controller for discrete user‑to‑reflector assignments and low‑level controllers that optimize continuous focal points via MAPPO under a CTDE scheme. Deterministic ray‑tracing tests show RSSI gains of up to 7.79 dB over centralized PPO baselines and robust performance with sub‑meter localization errors for multiple users and reflector arrays.
The paper introduces a provenance‑aware execution graph for long‑horizon LLM agents, defining influence distance (DI) as the shortest structural path from an untrusted source to a sensitive action. Compared to the traditional sequence distance (DT), DI is always less than or equal to DT, revealing a median gap of nine hops in 454 injection–sink pairs across multiple models and datasets. The study shows that most pairs exhibit a non‑zero gap, and a deterministic DI‑based gate can block attacks missed by a sequence‑only gate without extra benign blocking.
arXiv:2609.07247v1 Announce Type: new Abstract: Scaling the interaction horizon-the maximum number of environment interactions per episode-improves LLM agents on long-horizon tasks, and curriculum-ba...
The paper introduces the Agentic Adoption Index (AAI), a new measure of delegated exposure that captures whether workers actually commit tasks to AI within structured workflows. Using semantic embeddings of 888,000 agent skill specifications from GitHub and 18,000 O*NET task statements, the authors find that occupations with high delegation differ from those most vulnerable to pre-AI automation, that AAI correlates more with technical capability than with current LLM use, and that for lower‑educated occupations AAI rises with wages while it falls for higher‑educated, high‑earning workers. These patterns also appear in an independent corpus from the Manus Skills Marketplace.
AgentLens is a mobile GUI agent that adapts its visual communication during task execution, offering Full UI, Partial UI, and GenUI modalities. It uses a Virtual Display to allow background operation while selectively overlaying visual information. In a study with 21 participants, 85.7% preferred AgentLens, which also scored highest on usability and adoption intent.
An agentic framework called GW‑Eyes, powered by large language models, is introduced to autonomously associate gravitational‑wave (GW) signals with candidate electromagnetic (EM) counterparts. It integrates domain‑specific tools for tasks such as catalog management, skymap visualization, and rapid verification, while enabling natural‑language interaction to assist human experts. The framework leverages LLMs’ decision‑making and traceable reasoning to address the growing data‑analysis challenges of next‑generation GW and EM detectors.
AgentServeSim is a simulation framework designed to model the execution of large language model (LLM) agent programs, capturing cross‑turn key‑value (KV) state retention, successor turn release, and scheduling decisions. Unlike existing simulators that operate on request streams, AgentServeSim treats the entire agent program as a single unit of execution, using a Program Control Block, Program Orchestrator, Retention Plane, and Dispatch Plane to emulate realistic serving dynamics. Validation against real vLLM deployments on two GPU platforms shows mean job completion time errors below 5.5%, and the simulator enables automated policy search that improves mean JCT by up to 2.8% over hand‑written policies. whyItMatters":"The simulator provides a realistic, CPU‑based tool for evaluating and optimizing LLM agent serving policies, achieving high fidelity to real deployments and enabling measurable performance gains."
The paper investigates the use of a large language model (LLM) agent to automate the creation and execution of a federated learning workflow for medical image analysis. By generating a reviewable specification of constraints and acceptance criteria, the LLM produces an executable workflow, while a validation loop repairs failures and ensures conformance to the specification. Experiments on the FABRIC testbed demonstrate that the approach can detect silent errors—such as a run that trained 1,700 jobs on random tensors—that traditional failure-driven debugging would miss.