Tool use, function calling, orchestration and the protocols that let models act rather than only answer.
arXiv:2610.02349v1 Announce Type: cross Abstract: Inter-agent communication is central to Large Language Model Multi-Agent Systems (LLM-MAS), but it introduces an underexplored vulnerability: Agent-i...
arXiv:2610.02523v1 Announce Type: new Abstract: Developing cognitive models of algorithmic reasoning from behavioral data is a central problem in cognitive science that challenges current methods. Tr...
arXiv:2610.02869v1 Announce Type: cross Abstract: Autonomous penetration testing agents conduct multi-step attacks by continuously adapting their plans and actions to target responses. As a common de...
arXiv:2610.02769v1 Announce Type: cross Abstract: Language agents should draw on prior attempts and environmental feedback to improve subsequent decisions within the same task. However, providing add...
arXiv:2605.16358v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) are increasingly applied to real-world forecasting tasks, yet evaluating their true predictive capability remain...
arXiv:2604.10800v2 Announce Type: replace-cross Abstract: Learned classifiers deployed in agentic pipelines face a fundamental reliability problem: predictions are probabilistic inferences, not verif...
arXiv:2610.02765v1 Announce Type: cross Abstract: Monitoring planned driving trajectories requires accurately estimating the collision likelihood with actors whose motion is itself impacted by the eg...
arXiv:2610.03448v1 Announce Type: cross Abstract: LLM agents increasingly screen tool outputs with small prompt-injection detectors, and teams choose among detectors by their scores on public benchma...
arXiv:2604.05088v2 Announce Type: replace-cross Abstract: We propose ScalarFedLQR, a communication-efficient federated algorithm for model-free learning of a common policy in linear quadratic regulat...
arXiv:2610.02270v1 Announce Type: cross Abstract: Medical vision-language model (VLM) evaluation is sensitive to workflow design, prompting strategy, and benchmark construction, yet most studies trea...
MOF-VERIFY is a failure-aware agentic harness designed to improve hypothesis verification for metal‑organic frameworks (MOFs). It introduces a diagnostic benchmark with four task families—structural grounding, synthesis‑condition verification, evidence‑sufficiency verification, and MLIP‑based computational verification—to pinpoint failures in knowledge access, evidence acquisition, and reasoning. Guided by these diagnostics, MOF‑Verify addresses structural, literature, evidence‑sufficiency, and computational bottlenecks, achieving significant performance gains over direct inference and retrieval‑based baselines across multiple large language models.
arXiv:2606.20627v2 Announce Type: replace Abstract: Joint-Embedding Predictive Architectures (JEPAs) enable agents to plan in latent space by imagining the outcomes of candidate actions, yet task spe...
ULTRADISCOVERY is a benchmark that tests abductive exploration in an epistemically open, interconnected universe. It features five domains where agents must revise an initial theory and predict outcomes of unseen cross‑domain interventions, with a $2 imes 2$ design manipulating representation openness and evidence alignment. Experiments show that agents struggle to compose evidence into transferable representations, especially when the representation is open, and that even with aids exact predictions are rare without large budgets.
The article "How to Govern AI Agents" discusses strategies for overseeing a single AI agent and expanding those governance practices to manage a fleet of agents. It explores the challenges and solutions involved in scaling control mechanisms as the number of agents increases. The piece highlights practical approaches for ensuring reliable and ethical operation across multiple AI systems.
Simon Willison argues that pay‑by‑usage services and APIs should implement default hard budget caps that stop usage and return errors once a set monthly limit is reached, rather than merely sending warning emails. He contends that coding agents and personal agents often trigger costly API calls, and hard caps prevent unexpected large bills. Willison calls for a clear opt‑in checkbox to remove the cap, and notes that AWS has recently introduced spending limits to address this issue.
The article explores whether large language model (LLM) agents can discover new ideas by examining their creative potential. It frames this inquiry through the lens of creativity, aiming to assess how LLM agents might generate novel insights or solutions. The discussion is presented as a post on Towards Data Science.
The paper introduces a framework for routing financial decisions during system outages, using a local market‑impact response matrix to analyze expected squared price displacement. It shows that shared backups can create a concentration floor as primary endpoints increase, while balanced fallback risk diminishes with more surviving endpoints. For indivisible decisions, the authors derive exact second moments of randomized routing, propose conditional‑expectation rounding and a swap procedure, and demonstrate risk reductions of 6.57%–10.53% in synthetic tests and 3.30% in real API replay scenarios.
The paper introduces Density-Aware Reward Aggregation (DARA), a method that adjusts reward weighting in multi-reward reinforcement learning based on the density of active rewards within rollout batches. By applying an inverse-square-root density correction, DARA gives more weight to less frequently active rewards, enabling faster learning of targeted behaviors. Experiments on tool calling and mathematical reasoning demonstrate that DARA achieves comparable final performance while reducing training steps by up to 26% and 65% respectively.
The paper investigates how test‑time computation can enhance language models and at what cost, introducing the SELF‑POT benchmark to evaluate this across competition mathematics, competitive programming, and agentic workflows. SELF‑POT separates candidate coverage from final accuracy, tracks correctness transitions under revision, and measures protocol completion alongside task success. Using a unified budget rule, the study compares direct inference, parallel sampling, and self‑revision across five low‑cost reasoning models, revealing that selection rules and failure handling significantly influence gains and cost savings.