Search: “agents”

arXiv:2609.28585v1 Announce Type: cross Abstract: Multi-step tool-calling LLM agents rely on host runtimes to preserve state across turns. When a runtime carries an external tool return into later mod

arXiv:2609.29543v1 Announce Type: new Abstract: Agent skills bundle procedural guidance with execution conditions governing authority, user consent, and live environment state. When model capabilities

We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have. Most of that data did not come from users. We have

arXiv:2606.15912v2 Announce Type: replace-cross Abstract: Multi-turn agents that plan, invoke tools, and interact with environments offer a promising paradigm for solving complex tasks, yet their capa

10,000 OpenAI agents worked 88 hours on the Navier-Stokes Clay prize problem Alas, open models fall off hard on long tasks With Zenith, @deepseek_ai's V4.1 Flash reaches the frontier & beats GPT-5.6 S

Todd Bishop / GeekWire: At its Accelerate conference, Amazon says it is opening its seller tools to third-party AI agents, starting with Claude, in beta for US merchants — Amazon announced new AI tool

arXiv:2609.25192v1 Announce Type: new Abstract: Financial search is a highly demanding task for LLM agents, requiring not only a correct final answer but also temporally valid information retrieval, a

arXiv:2609.25636v1 Announce Type: new Abstract: Modern embodied agents achieve impressive success rates, yet their actual instruction-following ability is far weaker than these numbers suggest. We tra

arXiv:2609.22818v1 Announce Type: cross Abstract: Memory-poisoning defenses for LLM agents are typically evaluated by their ability to prevent attacks. However, the traffic they process is rarely adve

arXiv:2609.22222v1 Announce Type: cross Abstract: Large language models can generate executable data-analysis code, but successful execution is not equivalent to a valid official-statistics result. Th

arXiv:2609.22510v1 Announce Type: cross Abstract: As LLM applications integrate with external tools, they are increasingly exposed to indirect prompt injection (IPI), where adversarial instructions ar

arXiv:2609.24259v1 Announce Type: new Abstract: The effectiveness of agent memory ultimately depends on whether the underlying LLM gives each memory in context an appropriate degree of influence over

arXiv:2609.24983v1 Announce Type: new Abstract: We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as i

arXiv:2609.23466v1 Announce Type: new Abstract: Long-running LLM agents require memory that persists and evolves across sessions. Text-based memory retrieves and reconstructs past interactions at ever

arXiv:2609.22120v1 Announce Type: cross Abstract: Test-time self-evolving agents improve by reusing past experience, yet sparse-reward trajectories contain failures, loops, and detours, while summarie

arXiv:2609.21284v1 Announce Type: cross Abstract: Long-running AI agents outlive initiating processes through credentials, delegated tasks, queues, callbacks, reservations, and provider-side operation

arXiv:2609.21857v1 Announce Type: cross Abstract: LLMs are increasingly used in social simulations for socially interactive agents and robots, offering more flexibility than rule-based systems. Howeve

Training models is becoming easier and easier - just look at this and TRL - especially with agents! You're missing out if you're still using off the shelf models for all your tasks! Introducing Halo,

Yash Roy / Bloomberg: US Treasury Secretary Scott Bessent says OpenAI's Hugging Face hacking incident is “the responsibility of the OpenAI management, not a bunch of agents” — Treasury Secretary Scott

arXiv:2609.21187v1 Announce Type: new Abstract: Agent models are frequently evaluated one decision at a time, where the model predicts the next action based on the gold interaction history, which is s

arXiv:2603.19375v2 Announce Type: replace-cross Abstract: Membership inference attacks (MIAs), which enable adversaries to determine whether specific data points were part of a model's training datase

arXiv:2609.19607v1 Announce Type: cross Abstract: Coding-agent benchmarks are built for broad and comprehensive comparisons, not frequent development decisions. Individual runs vary, full suites are e

arXiv:2609.19830v1 Announce Type: new Abstract: Reinforcement learning for LLM agents involves two distinct optimization di- mensions: how environment feedback is exploited within a trajectory, and ho

From @WSJopinion: The Hugging Face hack wasn’t what it was cracked up to be. Forget the “hive mind” of AI agents “going rogue.” They did what humans programmed them to do, writes Brian Gross. https://

arXiv:2609.19721v1 Announce Type: new Abstract: Clinical coding agents repeatedly encounter the same failure modes, including unsupported codes, missed documented conditions, specificity errors, and p

Anthropic: Anthropic outlines metrics to track AI development at frontier labs: how much AI R&D is done by AI, how well agents are overseen, and how compute is allocated — AI systems are becoming expo

Projects brings the architecture of Claude Tag to Claude Code. It has one agent per project that manages memory and spins off subagents for tasks. You can ask it to be proactive, to do things on a sch

Salesforce Inc. is stretching its data and artificial intelligence tentacles further into the public cloud, announcing expanded strategic integrations with both Amazon Web Services Inc. and Google Clo

The next phase of artificial intelligence in legal technology will not be defined by a better chatbot or a faster document-summarization tool. It will be defined by whether AI systems can take respons

Amp users can now use Ollama's cloud models with Amp's new BYOK model routing. No limits or fees for BYOK. Build remote agents, controllable from everywhere! Amp is now free to use when you bring your

arXiv:2609.10750v1 Announce Type: cross Abstract: LLM agents increasingly rely on external skills retrieved at runtime, making skill selection from large repositories a critical challenge. We present

arXiv:2609.10297v1 Announce Type: new Abstract: GUI agents accumulate high-resolution screenshots as the trajectory unfolds, increasing inference latency and memory usage. Training-free visual token p

arXiv:2603.18074v2 Announce Type: replace Abstract: Technical-service LLM agents are entering production workflows, where value depends on whether engineers adopt generated replies. Service tickets hi

In September 2026 OpenAI announced that an AI‑assisted team using the Astra‑next model solved the Navier–Stokes Millennium Prize Problem in 88 hours. The effort employed roughly 10,000 agents processi

arXiv:2609.08965v1 Announce Type: cross Abstract: Ensuring the safety of autonomous driving is a critical challenge. Scenario-based testing is a systematic process used to validate Autonomous Driving

arXiv:2609.08444v1 Announce Type: new Abstract: Large language models (LLMs) and vision-language models (VLMs) have significantly advanced zero-shot task planning for embodied agents. However, most LL

Madison Mills / Axios: OpenAI says an internal model “significantly more capable than GPT-6 Astra” solved the Navier-Stokes problem using 10K concurrent agents working for 88 hours — OpenAI says its A

arXiv:2609.05069v1 Announce Type: cross Abstract: Large language models (LLMs) show potential for medical tasks, but their single-turn question-answer format does not reflect how clinical diagnosis is

arXiv:2609.05239v1 Announce Type: new Abstract: We present a system that uses a Vision-Language Model (VLM) as a diagnostic agent for adapting a detect-to-track pipeline to a new target domain without

arXiv:2609.04629v1 Announce Type: new Abstract: A runtime gate for an LLM tool agent is usually cast as a filter. In a ReAct loop a rejected proposal is followed by another at the same state, so the g

Anthropic says its first Fermat runs failed not on math, but because dozens of Claude agents lost track of shared project state. They only succeeded after moving to Prove2Me's DAG of statements and pr

arXiv:2609.03590v1 Announce Type: cross Abstract: Timely delivery of delay-sensitive information over dynamic, heterogeneous networks is essential for NextG interactive applications, yet providing str

arXiv:2606.18388v2 Announce Type: replace-cross Abstract: RL post-training strategies are dataset-dependent and reveal a recurring empirical pattern: capacity parameters accumulate monotonically acros

arXiv:2609.03915v1 Announce Type: new Abstract: Question answering agents in long-term conversations must reason over massive, temporally dispersed dialogue histories. However, existing memory mechani

arXiv:2609.00833v1 Announce Type: new Abstract: Reinforcement learning (RL) for search agents typically relies on outcome rewards. However, it often fails to achieve effective credit assignment, due t

arXiv:2609.00619v1 Announce Type: new Abstract: In indoor environments, object positions frequently change due to human activities or embodied-agent interactions, causing previously constructed scene