AI news archive.
4,966 articles filed under Agents, newest first, from the outlets listed on the sources page.
- arXiv cs.AIagentsAgentHabit: Characterizing Distinct Behaviors of Agents on Everyday Tasks
arXiv:2609.32795v1 Announce Type: new Abstract: Large language model (LLM) agents assist users with everyday tasks that can be completed in many reasonable ways. Even when their answers are useful, how agents carry out t…
- arXiv cs.AIagentsPlanGuard: A Guardrail for Multi-Step Plan Safety in Embodied Agents
arXiv:2609.32801v1 Announce Type: new Abstract: Embodied task planners may produce multi-step plans whose subtask dependencies and interactions with the environment create physical risks during execution. Yet existing sa…
- arXiv cs.AIagentsRe-derivability Decides What a Staged Agent Pipeline Recovers After an Upstream Fault
arXiv:2609.32802v1 Announce Type: new Abstract: One variable sets what an upstream fault costs a staged pipeline of language-model agents: re-derivability, how much of what a stage needs it can rebuild from the original…
- arXiv cs.AIagentsNutri-ATLAS: Embodied Agent for Tabulated Lookup and Assistance for Smarter nutrition
arXiv:2609.32803v1 Announce Type: new Abstract: Generative and Agentic IoT systems offer a promising foundation for digital healthcare applications that combine sensing, personalized reasoning, and autonomous interaction…
- arXiv cs.AIagentsDecision-Sufficient State Representations: Measuring and Reducing Write-Time Regret
arXiv:2609.32805v2 Announce Type: replace Abstract: Long tasks produce more history than an LLM agent can hold in its context, and more than it uses reliably even when the history fits. A growing line of work therefore h…
- arXiv cs.AIagentsImproving LLM Collaboration via Multi-Agent Preference Learning
arXiv:2609.32827v1 Announce Type: new Abstract: Several works have explored multi-agent reinforcement learning (MARL) in LLM collaboration. However, constructing reliable rewards is difficult in practice, as complete and…
- arXiv cs.AIagentsFinancialAuditBench: Benchmark Construction under Differential Privacy Using Real-World Priors
arXiv:2609.32835v1 Announce Type: new Abstract: As AI agents are becoming widely adopted in the financial services industry, careful measurement is essential to understand where they can be reliably deployed and where ov…
- arXiv cs.AIagentsCounterfactual Self-Evolving Agents for Evidence-Grounded Reasoning
arXiv:2609.32870v1 Announce Type: new Abstract: Self-play proposer--solver methods improve reasoning by generating tasks and learning from verified solutions. However, for evidence-identifiable tasks, where case-specific…
- arXiv cs.AIagentsPlanner-as-Router: Joint Plan-Time Model Routing for Cost-Efficient Multi-Agent Workflows
arXiv:2609.32917v1 Announce Type: new Abstract: Running large language model (LLM) agents in production gets expensive fast. A frontier model (the largest, most capable tier) is accurate but can cost 25 times what a smal…
- arXiv cs.AIagentsRelic: From Multi-Agent Collaboration to Persistent Organizational Capability
arXiv:2609.32965v2 Announce Type: replace Abstract: Multiple agents may often conflict in an organization: for example, one coding agent changes an interface in a repository, but another continues to develop on the old v…
- arXiv cs.AIagentsCertified Long-Horizon Code Agent Evolution via Validation-Gated Skill Optimization
arXiv:2609.32990v1 Announce Type: new Abstract: Long horizon agent self-evolution without model weight updates is essential for enabling deployed agents to accumulate reusable skills and improve over time. Prior self-evo…
- arXiv cs.AIagentsX-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization
arXiv:2609.32993v1 Announce Type: new Abstract: Multi-step agents are trained on flat action streams: SFT and RLVR weight every token uniformly and ignore the sub-procedures that recur across tasks, the hierarchy that le…
- arXiv cs.AIagentsWhen Pair Count Is Not the Sample Size: What All-Pairs Agent Comparisons Estimate
arXiv:2609.33012v1 Announce Type: new Abstract: When an agent benchmark compares every pair of leaderboard entries, the number of comparisons can look much larger than the independent evidence behind them: A versus B and…
- arXiv cs.AIagentsThe Epistemics of Agent Memory: Measuring, and Governing, the Consolidation Decision in Long-Horizon LLM Agents
arXiv:2609.33013v1 Announce Type: new Abstract: Long-horizon LLM agents must convert accumulated experience into durable memory, deciding what to keep, compress, abstract into reusable skills and rules, or forget. We rep…
- arXiv cs.AIagentsTrust and Task Completion in the World of Consumer AI Agents
arXiv:2609.33017v1 Announce Type: new Abstract: Action agents do things for people. They send email, spend money, and call businesses while the user is busy with something else, so a mistake can turn into an action befor…
- arXiv cs.AIagentsSRE-Marathon: A Continuous, Change-Driven Benchmark for Autonomous Site Reliability Agents
arXiv:2609.33023v1 Announce Type: new Abstract: Benchmarks for site reliability engineering (SRE) agents are typically episodic: one fault is injected, the agent receives an incident task, and its response is scored. Pro…
- arXiv cs.AIagentsAgent Safety From Within: Detecting Harmful Trajectories from LLM Internal States
arXiv:2609.33039v1 Announce Type: new Abstract: Language model agents can now perform sophisticated sequences of actions via tools and harnesses, which has increased the scope of the damage they can cause. Guard models,…
- arXiv cs.AIagentsCompositional Safety Failures in Harness Evolution: Identification and Runtime Monitoring
arXiv:2609.33123v3 Announce Type: replace Abstract: Self-evolving agent harnesses continually update persistent components such as memory, prompts, skills, and tools. We call this process harness evolution. However, such…
- arXiv cs.AIagentsOn Device Agentic Operation Caches -- Classifier-Centric NL-to-Action Generation
arXiv:2609.33141v1 Announce Type: new Abstract: Agentic AI is increasingly being embedded in software applications to provide natural language interfaces to features and functionality. In most cases these agents are powe…
- arXiv cs.AIagentsLiteEvo: Automated, Cost-Efficient Harness Evolution for Generalization to Unseen Tasks
arXiv:2609.33146v1 Announce Type: new Abstract: An LLM agent is defined by two things: the weights inside its model and the harness of components assembled around it. Harnesses are still handcrafted, and HarnessX, which…
- arXiv cs.AIagentsWorldAgent: Verification-Guided Agentic Physical World Construction
arXiv:2609.33208v1 Announce Type: new Abstract: Constructing complex physical worlds from language requires coordinating extensive 3D environments, detailed structures and objects at different spatial scales, and interac…
- arXiv cs.AIagentsCodeSkill: Latent Skill Abstraction for Long-Horizon Code Agents
arXiv:2609.33243v2 Announce Type: replace Abstract: Code agents require long-horizon decision-making over complex interaction trajectories. However, existing reinforcement learning (RL) approaches typically optimize beha…
- arXiv cs.AIagentsActiveMem: Dynamic Latent Memory Trees for Long-Horizon Agents
arXiv:2609.33244v2 Announce Type: replace Abstract: Large Language Model (LLM) agents increasingly rely on external memory to support long-horizon reasoning and decision making. Existing memory systems typically retrieve…
- arXiv cs.AIagentsCORTEX: A Verified Experience Layer for Generalist Agents
arXiv:2609.33260v1 Announce Type: new Abstract: An agent can solve a task today and face the same task under new facts, tools, or governing knowledge tomorrow. Most agent systems can retrieve relevant text or recall prio…
- arXiv cs.AIagentsLSTMem: Hierarchical Long Short-Term Online Memory for Large Language Models
arXiv:2609.33268v2 Announce Type: replace Abstract: Large language models increasingly serve as long-horizon assistants and agents, where they must both accumulate information across interactions and make the relevant pa…
- arXiv cs.AIagentsLearning to Sell: Reinforcement Learning for Strategic Large Language Model Agents in Multi-Product Markets
arXiv:2609.33289v2 Announce Type: replace Abstract: Autonomous large language model (LLM) agents operating in multi-product markets must make sequential decisions under information asymmetry and resource constraints. We…
- arXiv cs.AIagentsTraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces
arXiv:2609.33295v1 Announce Type: new Abstract: An agent can complete a task while exhibiting undesirable behavior during execution. Developers need tests for the specific behaviors encountered in deployment, beyond fixe…
- arXiv cs.AIagentsAgentic Multi-Turn Reasoning: A Fairness Approach
arXiv:2609.33323v1 Announce Type: new Abstract: Recent advances in Large Language Models (LLMs) have enabled agentic systems capable of solving complex tasks through multi-turn planning, tool use, verification, and memor…
- arXiv cs.AIagentsANTMAN: Adaptive Need Tracking for Multi-Agent Navigation in Large Information Spaces
arXiv:2609.33326v1 Announce Type: new Abstract: Information-seeking agents increasingly operate over information spaces that are too large to process exhaustively. Yet many multi-agent systems organize computation around…
- arXiv cs.AIagentsLong-Horizon Analog Design Bench: Benchmarking Agents on Hours-Long Analog and Mixed-Signal Circuit Design Tasks
arXiv:2609.33356v1 Announce Type: new Abstract: Coding agents now sustain hours-long, tool-driven loops, yet their ability to carry long-horizon analog and mixed-signal circuits to electrical specification remains unmeas…
- arXiv cs.AIagentsDISCERN: Can AI Agents Work Like Scientists and Guide Discovery?
arXiv:2609.33357v2 Announce Type: replace Abstract: Reliable automated research requires agents to vet data, verify analyses, and generate hypotheses grounded in trustworthy evidence, potentially reducing routine scienti…
- arXiv cs.AIagentsAPEX: An Extensible Model for Agent-Assisted Production Scheduling
arXiv:2609.33430v1 Announce Type: new Abstract: Production scheduling requires realistic models that reflect operational constraints and efficient methods that balance competing goals. Putting these methods into use also…
- arXiv cs.AIagentsRaven: The Harness of Harnesses for Composable Agentic Intelligence
arXiv:2609.33439v1 Announce Type: new Abstract: As large language models advance, AI agents are moving beyond isolated, domain-specific tasks toward long-horizon, cross-domain workflows. This transition exposes two chall…
- arXiv cs.AIagentsLiveOption: Evaluating LLM Agents in Structured Option Trading with Nonlinear Payoffs
arXiv:2609.33470v2 Announce Type: replace Abstract: Large language models (LLMs) and multi-agent systems (MAS) have shown promise in financial decision-making, yet existing evaluations focus on equity trading and primari…
- arXiv cs.AIagentsFedMHAR: Federated Multimodal Human Activity Recognition using Multi-Agent Reinforcement Learning
arXiv:2609.33492v2 Announce Type: replace Abstract: Human Activity Recognition (HAR) from heterogeneous wearable sensors is fundamental to the Internet of Health Things (IoHT), supporting rehabilitation, elderly care, an…
- arXiv cs.AIagentsWhat Happens During Autonomous Deep Research After the User Steps Away?
arXiv:2609.33509v1 Announce Type: new Abstract: In autonomous deep research, a user provides a task and relevant background, then leaves the agent to conduct an extended investigation without further human intervention.…
- arXiv cs.AIagentsEverMine: Dissecting the Self-Evolution of Research Capabilities in Long-Horizon Alpha Research
arXiv:2609.33524v1 Announce Type: new Abstract: Self-evolving agents aim to turn research feedback into reusable skills, tools, and research rules. Whether these accumulated capabilities continue to improve later researc…
- arXiv cs.AIagentsDr. Free: You Don't Need Difficulty Rewards for Self-Evolving Search Agents
arXiv:2609.33565v1 Announce Type: new Abstract: A central limitation of current data-free self-evolution methods for training search agents is their reliance on difficulty-based proposer rewards. These methods reward a p…
- arXiv cs.AIagentsParaAgent: Reinforcing Parallel Acting in Open-World Tool Environments
arXiv:2609.33618v1 Announce Type: new Abstract: Language model agents are increasingly deployed in open-world tool environments, which require balancing exploring unknown capabilities and exploiting known ones. Existing…
- arXiv cs.AIagentsTrajectory Unlearning on LLM-based Agents
arXiv:2609.33639v1 Announce Type: new Abstract: Existing large language model (LLM) unlearning has focused primarily on removing specific knowledge, such as harmful facts, private data, or copyrighted content. However, a…
- arXiv cs.AIagentsProbe to Act: Elevating Browser-Use Agent via Active Visual Probing
arXiv:2609.33646v1 Announce Type: new Abstract: Browser-use agents require seamless alignment between structured web metadata and visual information, while preserving relevant context across long interactions. Existing i…
- arXiv cs.AIagentsAgentBoundary: Counterfactual Evaluation of Safety in Tool-Using LLM Agents
arXiv:2609.33658v1 Announce Type: new Abstract: Safety alignment for large language models (LLMs) in conversational settings is largely framed around whether to answer or refuse a request. In agentic settings, however, t…
- arXiv cs.AIagentsCompoWorld: Compositional Environment Scaling for General Agents
arXiv:2609.33665v2 Announce Type: replace Abstract: Automatically generated environments provide a scalable source of interaction data for training general agents. However, existing approaches mainly generate tasks withi…
- arXiv cs.AIagentsAuditing Agent Actions through Query-Conditioned Attribution
arXiv:2609.33676v2 Announce Type: replace Abstract: LLM agents increasingly take consequential actions through interactions with users, policies, and external tools. Auditing these agents requires automated attribution o…
- arXiv cs.AIagentsSWE-Game: Can Coding Agents Build the Games We Want?
arXiv:2609.33678v3 Announce Type: replace Abstract: We introduce SWE-Game, a benchmark of 247 tasks grounded in 41 executable reference Godot games spanning 13 gameplay categories in 2D and 3D. Five task types cover deve…
- arXiv cs.AIagentsSelf-Designed Evaluators and Warm Memory for Long-Horizon Agents
arXiv:2609.33717v1 Announce Type: new Abstract: A tool-using language-model agent deployed over a long stream of tasks receives no reward, so it cannot tell whether it succeeded, cannot safely retry, and cannot label the…
- arXiv cs.AIagentsHTN Planning as a Coordination Layer for Multi-Server MCP Tool Orchestration
arXiv:2609.33731v1 Announce Type: new Abstract: The Model Context Protocol (MCP) isolates servers by design: only the host can orchestrate cross-server workflows. When the host is a large language model, the resulting or…
- arXiv cs.AIagentsSkill2Env: Capability-Oriented Environment Synthesis from Skills for General Agents
arXiv:2609.33772v2 Announce Type: replace Abstract: Executable environments are critical for post-training agents on tasks that require tool use and multi-step interaction, but constructing executable tasks together with…
- arXiv cs.AIagentsLearning Strategies to Break Judges
arXiv:2609.33773v1 Announce Type: new Abstract: As AI agents surpass human performance, it becomes exceedingly hard for system designers to evaluate them directly and understand their failure modes. Consequently, agents…
- arXiv cs.AIagentsEvidence-Inference Reconstruction: When The Evidence Is Recalled But The Reasoning Goes Wrong
arXiv:2609.33778v1 Announce Type: new Abstract: Modern multi-hop LLM agents are equipped with built-in mechanisms to detect errors in intermediate reasoning steps. Such errors trigger corrective actions from these agents…