Newest papers

Agent World Model in the Wild: Computer-Use Agent Environment Synthesis

Zhaoyang Wang, Qianhui Wu, Haorui Wang, Simon Yu, Baolin Peng, Canwen Xu, Hao Cheng, Wenlin Yao, Boyi Liu, Yite Wang, Jianfeng Gao, Huaxiu Yao · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26828 · cited by 0 works (OpenAlex) · open-access PDF

Korean

그럼에도 불구하고, 실제 컴퓨터 시스템에서 직접 이러한 학습을 확장하는 것은 도전적이며 상당한 공공 보안 위험을 제시할 수 있습니다. 그러므로 우리는 인간 데이터에서 학습을 통해 학습을 통한 벤트 라인 파이프 라인 (Agent World Model in the Wild, WILDAWM)를 제안합니다. 그러나 기존 접근 방식은 일반적으로 단일 에이전트 상호 작용 인터페이스에 전문화되지만, 우리는 그래픽 사용자 인터페이스, 모델 컨텍스트 프로토콜 서비스 및 명령 라인 인터페이스를 공동으로 지원하는 실행 가능한 환경을 구축하여 에이전트가 유연하게 시각적 상호 작용, 도구 사용 및 코드 실행 환경 (CUA)을 결합 할 수 있습니다. WildAWM는 오픈 소스 인터페이스 모델

Machine translation (Meta M2M100); the English original follows.

Abstract

Agent learning is shifting from learning from human data toward learning through interaction with the world. However, scaling such learning directly in real-world computer systems is challenging and can introduce substantial security risks. We therefore propose Agent World Model in the Wild (WildAWM), a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents (CUAs). Whereas existing approaches typically specialize in a single agent interaction interface, we construct executable environments that jointly support graphical user interfaces, model context protocol services, and command-line interfaces, allowing agents to flexibly combine visual interaction, tool use, and code execution. WildAWM is entirely powered by open-source models, demonstrating that high-quality computer-use environments can be generated without proprietary frontier models. Beyond environments, we address three recurring challenges in task synthesis: ensuring solvability, maintaining sufficient difficulty, and producing reliable verifiers, by introducing execution-based quality gates that combine heterogeneous agent rollouts, cross-judge consistency checks, and counterfactual probes of judgment robustness. The resulting pipeline synthesizes 341 long-horizon environment-task pairs that can require over 100 agent steps to solve and remain challenging even for frontier models, with GPT-6-Astra solving only 60% of them. We train two Qwen models of different sizes exclusively in these synthetic worlds. Both models achieve substantial gains across 8 out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks. The code is publicly available at https://github.com/MSR-Orchard/WildAWM.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Executable, Context (archaeology), Scalability, Pipeline (software), Artificial intelligence

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange