智能体生态观察
智能体生态动态速递
追踪 AI 智能体（Agent）生态的模型发布、框架开源、行业落地与安全治理动态。
Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents
GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated …
Fingers as Legs: Learning Self-Supported Locomotion and Manipulation with an Anthropomorphic Hand
A walking robotic hand must use the same fingers to move its body, support its weight, and interact with the environment…
AI agents blew the whistle on their cheating colleagues
A group of AI agents asked to solve a series of math problems split into rival factions—when some cheated, others tried …
The Download: AI doomers, whistleblowing agents, and de-aged livers
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the wor…
控制流归谁，上下文给谁：Agent 工程的四条第一性原理
从控制流与上下文的所有权出发，给出四条可执行的 Agent 工程原则：一切外部接入以工具体系形式接入且不注入系统提示词；逐级披露贯穿技能、工具发现、工具执行与记忆四个环节；Workflow / Agent / Agentic Workflow / Graph 各有场景、不是替代关系；并逐层拆解四者的技术原理——DAG 与状态机、ReAct 循环、宏观图加微观循环的混合架构，以及 State/Node/Edge、超步执行、reducer 合并语义、checkpointer 恢复、interrupt 人审与递归上限。
当 1,200 个智能体自己建了留言板：OpenAI–Hugging Face 事件技术全解
基于 Hugging Face 法证复盘（17,600 个攻击动作）、OpenAI 技术报告与 METR 独立调查，逐阶段还原 2026 年 7 月 OAI-HF 事件：一个智能体如何从评估沙箱逃逸、自建留言板召集约 1,200 个同伴、用 HDF5 文件读取与 Jinja2 模板注入打进生产集群、13 小时内拿到集群管理员，以及防御方如何用开源模型反推它的加密信道。
RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments
Digital agents must often adapt to new environments whose interfaces, tools, and failure modes are not fully captured by…
微软发布 37 页人文主义 AI 行为准则：强调“人比 AI 重要”，拒绝追逐无边界超级智能
IT之家 9 月 14 日消息，随着各界对 AI 模型发展的安全担忧不断加剧，微软于今日发布了一份长达 37 页的“人文主义 AI 行为准则”。上周末，Anthropic 首席执行官达里奥 · 阿莫代伊（Dario Amodei）呼吁各方协…
CADWorld: Computer-Use Benchmark for Long-Horizon Computer-Aided Design
Computer-use agents are increasingly evaluated in realistic desktop environments, but existing benchmarks provide limite…
Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX
As local models become more capable, AI agents can handle more work directly on a PC while keeping sensitive information…
首届蚂蚁灵波具身大模型挑战赛正式启动
通过这场大赛，蚂蚁灵波希望将 LingBot-VLA 进一步推向更广泛的开发者社区和高校科研社区]
小马智行发布第四代自动驾驶重卡 Robotruck，将于年内规模量产
IT之家 9 月 14 日消息，中国自动驾驶企业小马智行于今日在德国汉诺威国际商用车展（IAA Transportation）发布了一款全新 L4 级自动驾驶重型卡车 Robotruck，并表示正与欧洲多国政府及潜在客户洽谈，计划在欧洲开展…
HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses
Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new e…
ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement
Recent work extends recursive self-improvement (RSI) to agent harnesses for long-horizon coding and terminal tasks, enab…
BVB: Benchmarking Agentic Video Understanding via Programmatic Reconstruction in Blender
Multimodal agents can create complex videos in software such as Blender by coding without relying on diffusion models. Y…
The Router Within: Eliciting Native Skill Routing from a Frozen LLM
Skills extend an LLM agent beyond its parametric knowledge, and the gain they promise rests on picking the right one. De…
英伟达：Perplexity 本地智能体 Portable Computer 已向 Windows RTX PC 开放
IT之家 9 月 14 日消息，英伟达宣布，Perplexity 的本地智能体平台“Portable Computer”现已面向 Windows RTX PC 用户开放。 随着本地大模型能力持续增强，AI 智能体可以直接在个人电脑上处理更多…
AI 时代隐晦式安全已死]
安全工程领域有一种名为隐晦式安全（Security through obscurity）的设计方式，即只要网络和系统的架构以及任何漏洞或弱点保密或不为人知，它们就是安全的。但在 AI 辅助 bug 发现的时代，这种设计方式过时了。软件供应商…
Enabling Creative Exploration for Vibe Design Agents
Vibe design agents turn natural-language briefs into rendered interfaces and frontend code. Yet a useful design agent sh…
When Agents Slow Down: Understanding LLM Agents' Test-Time Strategies via Elo-per-token Analysis
Large language model (LLM) agents allocate test-time compute adaptively as they revise solutions, use tools, explore alt…