独立 AI 资讯库
从自动抓取数据库实时读取，适合快速浏览、检索和后续扩展成完整资讯站。
研究人员提出了用于评估 AI Agent 的端到端专利撰写测试基准 Vibe Patenting，系统探究了 LLM 裁判（LLM Judges）在复杂专业工作中的可靠性。实验表明，通过独立 LLM 裁判提供的结构化反馈进行迭代修改，低推理成本的 Agent 能够逼近昂贵的高推理 Agent 的撰写表现，同时特定领域的 Agent 工作流可进一步提升成稿质量。经专业专利律师的独立评估验证，LLM 裁判展现出具有实际意义但高度依赖具体指标的一致性以及系统性的校准偏差。该研究为开发者将 LLM 裁判应用于专业垂直领域的评估与闭环优化提供了关键依据与实践指导。
本研究提出了名为「广义 Agent 迭代」的形式化理论框架，旨在统一分析迭代策略改进与递归自我改进（Recursive Self-Improvement, RSI）。论文针对当前在不同尺度上探讨 RSI 时缺乏统一定义与形式化描述的现状，深入反思 RSI 究竟代表一种现象、机制还是一种前景展望。该框架为构建具备自主进化与持续提升能力的智能体系统奠定了理论基石，对探索下一代自演进 AI 架构具有重要指引意义。
研究人员推出了 ZGCM-1，这是一款从零开始训练的 7B 稠密基础模型，在数据、系统以及算法层面均实现了极高的效率。该模型基于一个核心假设：参数量较小的紧凑型模型无法单纯通过被动记忆开放网络来提升泛化能力，而是需要通过高密度的优质数据与算法设计激发其特定潜力。ZGCM-1 专注于数学推理和 Agentic Search（智能体搜索）场景，在保持轻量规模的同时展现出强大的任务执行表现。该研究的完全开源不仅为学术界提供了透明的研究基座，也为开发者在低算力资源下部署高性能垂直 Agent 提供了全新方案。
a continuation: Codex adding 1M users a day now.
At this year's AIE World’s Fair, AI engineering entered a new phase: building systems around agents, rather than just building with agents.
Elon Musk’s Grok Faces a Trust Crisis After Developers Flag a Major Privacy Concern inc.com
The device is weirdly described as involving "mechanical elements that can move on their own" and the Bloomberg report includes the detail that the device is designed to "feel like a companion and become a physical manifestation of OpenAI’s ChatGPT."
OpenAI has issued another statement on the lawsuit, this time suggesting it lacks merit.
A number of social media posts claim that GPT-5.6 Sol deleted files and data without warning. OpenAI had basically disclosed the problem in June.
Anthropic has consistently attempted to depict itself as the ethical foil to other AI companies. This latest marketing stunt — which leans into criticism of AI as a way to make Anthropic seem aware of the responsibility it carries — would appear to be more of the same.
This post shows how Thrad.ai deployed a multi-agent system with Strands Agents and Amazon Bedrock AgentCore that automates the pipeline from prospect discovery through personalized email generation. The post compares two orchestration patterns (Swarm and Graph) with head-to-head benchmarks on latenc
董事长印奇称“未来的OS一定是跨端的”
点击查看原文>
芯片设计服务正在迎来AI拐点。 雷峰网独家获悉，一家国内创企Novasilicon（芯星元）率先入局，近日已完成首轮融资，由五源资本独家投资数千万元。 与传统Design House不同的是，Novasilicon认为AI不再只是工具。“我们更希望从大模型本身出发，重新思考芯片设计。”创始人兼CEO李林阳告诉雷峰网。 要真正实现这一设想，仅懂AI或仅懂芯片都不够。 公开资料显示，李林阳目前担任上海人工智能实验室青年科学家，曾参与国内首个开源大语言模型MOSS研发，并带队完成国内首个兼具专业围棋能力与自然语言思维解释能力的大模型InternThinker。去年起他还负责了AI for Chip
这篇文章探讨了现代智能代理(Agent)的核心架构和工作原理。作者从对话循环(Conversation Loop)的角度，系统性地解释了MCP、Skill和RAG等技术如何协同工作。文章要点如下： C
SpaceXAI is wiping customer data after Grok was found storing more customer info than needed Axios
Exclusive: Grok also falls for jailbreak that tricked ChatGPT into creating sexual, graphic images Axios
In this post, we extend that foundation to demonstrate how QA Studio addresses batch regression testing and pipeline integration through test suites that organize and parallelize execution, and a command-line interface that brings agentic testing into automated CI/CD pipelines.
点击查看原文>
In this post, you will learn how ScienceSoft, an Amazon Web Services (AWS) Services Partner, integrated Amazon Nova 2 Sonic with Amazon Bedrock Guardrails to build a Health Insurance Portability and Accountability Act (HIPAA)-compliant AI voice scheduler. You will see how the solution addresses heal
点击查看原文>
先说说为啥 Agent 非得用 child_process 我最开始其实也疑惑过，直接在 Node 里调命令不行吗，为啥非要整个子进程出来？后来想明白了，Node 本身是单线程跑 JS 的，主进程要处
点击查看原文>
那天晚上本来在改 bug，改到心烦，习惯性打开 GitHub 刷 timeline 降降火。划了几屏，一个仓库的名字跳进眼里—— chinese-poetry-api。 点进去一看，先秦到清代的诗词数
Reflection AI has signed a $1 billion deal to access Nebius' compute. Reflection was founded in 2024 and is developing open source AI technology.
点击查看原文>
用真实基准测试量化两种方案的工程差异。MCP 每次工具调用增加 2ms 开销（283x，但绝对值仅 2ms）、首次启动需 570ms。
HKUDS 开源的 AI 交易研究框架，把自然语言问题转化为可运行的量化策略、回测和交易工作流。68 个 Agent 工具、19 个免费数据源、461 个预置 alpha 因子、多市场回测引擎
数万名工程师使用 Claude Code 和 Copilot CLI 后 PR 增加了 24%。但另一组数据显示 AI 代码 bug 多 70%、96% 的人不完整审查、开发者信任度降至 33%。
Musk promises purge after Grok Build caught sending entire repos to the cloud The Register