{
 "version": "https://jsonfeed.org/version/1.1",
 "title": "WAKENEX — AI Frontier: latest",
 "home_page_url": "https://wakenex.com/topics/ai/",
 "feed_url": "https://wakenex.com/topics/ai/latest.json",
 "description": "Follow AI from the primary source, then ask what it actually changes.",
 "language": "en",
 "items": [
  {
   "id": "b26f70c790034ce0",
   "url": "https://github.blog/changelog/2026-10-26-github-copilot-assisted-permissions-generally-available",
   "external_url": "https://github.blog/changelog/2026-10-26-github-copilot-assisted-permissions-generally-available",
   "title": "GitHub Copilot assisted permissions generally available",
   "content_text": "Assisted permissions are now generally available in GitHub Copilot CLI, GitHub Copilot app, and Visual Studio Code. When Copilot requests permission to run a command, access a file system path,… The post GitHub Copilot…",
   "language": "en",
   "_wakenex": {
    "source": "github-copilot",
    "source_name": "GitHub Changelog · Copilot",
    "group": "work",
    "why_en": "Copilot now requests permission before running commands or accessing files, limiting silent side-effects.",
    "why_zh": "Copilot执行命令前会请求权限，可据此检查自己的开发环境是否启用了该约束。",
    "title_zh": "GitHub Copilot 辅助权限功能正式全面可用",
    "summary_zh": "辅助权限功能现已在 GitHub Copilot CLI、GitHub Copilot 应用及 Visual Studio Code 中正式全面可用。当 Copilot 请求执行命令、访问文件系统路径等权限时……",
    "original_language": "en"
   },
   "date_published": "2026-10-26T22:31:30Z"
  },
  {
   "id": "3e66020cfa4cc99c",
   "url": "https://epochai.substack.com/p/can-ai-automate-epoch",
   "external_url": "https://epochai.substack.com/p/can-ai-automate-epoch",
   "title": "We tried to automate Epoch",
   "content_text": "Top models complete about two-thirds of our rubric on real work tasks.",
   "language": "en",
   "_wakenex": {
    "source": "epoch-ai",
    "source_name": "Epoch AI · Gradient Updates",
    "group": "compute",
    "why_en": "Top models finish ~two-thirds of real work tasks; calibrate how much to trust AI output.",
    "why_zh": "顶级模型完成约三分之二的真实任务，帮助校准对AI输出的信任程度。",
    "title_zh": "我们尝试将 Epoch 自动化",
    "summary_zh": "顶尖模型在真实工作任务评分标准上完成了约三分之二。",
    "original_language": "en"
   },
   "date_published": "2026-10-26T16:33:24Z"
  },
  {
   "id": "3957ef832b45a4be",
   "url": "https://www.infoq.cn/article/O7lcxqKD2avRPbNoMmtg",
   "external_url": "https://www.infoq.cn/article/O7lcxqKD2avRPbNoMmtg",
   "title": "Behind 30 Agents: How Huajiao Built a Sustainably Evolving Enterprise AI Capability",
   "content_text": "Behind 30 Agents: How Huajiao Built a Sustainably Evolving Enterprise AI Capability",
   "language": "en",
   "_wakenex": {
    "source": "infoq-cn",
    "source_name": "InfoQ China",
    "group": "coverage",
    "why_en": "A real enterprise scaling 30 agents reveals what organizational structure actually holds.",
    "why_zh": "真实企业部署30个智能体的结构经验，可用于检验自身AI能力建设路径。",
    "title_zh": "30 个 Agent 背后：花椒如何建设可持续演进的企业 AI 能力？",
    "summary_zh": "",
    "original_language": "zh"
   },
   "date_published": "2026-10-26T15:05:27Z"
  },
  {
   "id": "af90007c20a665ff",
   "url": "https://www.infoq.cn/article/NrUkJExfuuFOICsjVYbb",
   "external_url": "https://www.infoq.cn/article/NrUkJExfuuFOICsjVYbb",
   "title": "Models Are Easy to Choose, Search Is Hard to Build: Two Years of Developing an Investment Research Agent",
   "content_text": "Models Are Easy to Choose, Search Is Hard to Build: Two Years of Developing an Investment Research Agent",
   "language": "en",
   "_wakenex": {
    "source": "infoq-cn",
    "source_name": "InfoQ China",
    "group": "coverage",
    "why_en": "Two years of production shows search quality, not model choice, limits research agents.",
    "why_zh": "两年实践表明，检索质量而非模型选择才是研究类智能体的核心瓶颈。",
    "title_zh": "模型好选，搜索难做：一个投研Agent的两年开发历程",
    "summary_zh": "",
    "original_language": "zh"
   },
   "date_published": "2026-10-26T14:57:35Z"
  },
  {
   "id": "8eb6001a836a6e50",
   "url": "https://openai.com/index/building-scientific-understanding-with-academic-communities",
   "external_url": "https://openai.com/index/building-scientific-understanding-with-academic-communities",
   "title": "Working with academic communities to build scientific understanding",
   "content_text": "OpenAI shares model results on the Kakeya conjecture, Exponential Time Hypothesis, and quantum chromodynamics—and how researchers can verify and build on them.",
   "language": "en",
   "_wakenex": {
    "source": "openai-news",
    "source_name": "OpenAI News",
    "group": "releases",
    "why_en": "Sharing verifiable model outputs on open problems shows how to check AI reasoning in formal domains.",
    "why_zh": "公开可验证的模型推理结果，提供了在严谨领域核查AI输出的方法。",
    "title_zh": "与学术界携手共建科学认知",
    "summary_zh": "OpenAI分享了模型在Kakeya猜想、指数时间假设及量子色动力学方面的研究结果，以及研究人员如何对其进行验证和深化。",
    "original_language": "en"
   },
   "date_published": "2026-10-26T14:30:00Z"
  },
  {
   "id": "68a3a72cda7cccff",
   "url": "https://importai.substack.com/p/import-ai-478-math-for-alignment",
   "external_url": "https://importai.substack.com/p/import-ai-478-math-for-alignment",
   "title": "Import AI 478: Math for alignment; ProgramBench; and Mirendil shows what RSI means",
   "content_text": "What is “night” to a machine?",
   "language": "en",
   "_wakenex": {
    "source": "import-ai",
    "source_name": "Import AI (Jack Clark)",
    "group": "research",
    "why_en": "Some human concepts may lack grounding in AI; check which ones your workflow assumes.",
    "why_zh": "部分人类概念对AI而言缺乏真实基础，使用前应识别哪些概念被模型误解。",
    "title_zh": "Import AI 478：对齐研究中的数学；ProgramBench；以及 Mirendil 展示 RSI 的含义",
    "summary_zh": "对机器而言，\"夜晚\"是什么？",
    "original_language": "en"
   },
   "date_published": "2026-10-26T12:31:52Z"
  },
  {
   "id": "27334100e818b985",
   "url": "https://openai.com/index/chatgpt-expanding-world-of-work",
   "external_url": "https://openai.com/index/chatgpt-expanding-world-of-work",
   "title": "ChatGPT’s expanding world of work",
   "content_text": "Explore OpenAI Signals data on 165 work activities across nearly 100 countries, revealing how people use ChatGPT for work and how usage changes over time.",
   "language": "en",
   "_wakenex": {
    "source": "openai-news",
    "source_name": "OpenAI News",
    "group": "releases",
    "why_en": "Usage data across 165 tasks lets you benchmark which work activities AI reliably handles.",
    "why_zh": "165项工作任务的使用数据，帮助判断哪些工作AI可以稳定胜任。",
    "title_zh": "ChatGPT不断拓展的工作应用场景",
    "summary_zh": "探索OpenAI Signals数据，涵盖近100个国家的165项工作活动，揭示人们如何将ChatGPT用于工作，以及使用方式如何随时间变化。",
    "original_language": "en"
   },
   "date_published": "2026-10-26T10:00:00Z"
  },
  {
   "id": "cc882ac7929ae4b8",
   "url": "https://www.qbitai.com/2026/10/505529.html",
   "external_url": "https://www.qbitai.com/2026/10/505529.html",
   "title": "OpenAI Begins Air-Gapped Training of Its Most Powerful Model: \"For Safety Reasons\"",
   "content_text": "Are models starting to use GPU cache to chat?",
   "language": "en",
   "_wakenex": {
    "source": "qbitai",
    "source_name": "QbitAI",
    "group": "coverage",
    "why_en": "Air-gapped training isolates powerful models from external data during training for safety containment.",
    "why_zh": "气隙隔离训练是一种安全机制，可据此追问任何高能力模型的训练环境是否做了类似隔离。",
    "title_zh": "OpenAI开始断网训练最强模型：“安全起见”",
    "summary_zh": "模型开始利用GPU缓存聊天？",
    "original_language": "zh"
   },
   "date_published": "2026-10-26T09:37:57Z"
  },
  {
   "id": "a9ab3f431f5a11b4",
   "url": "https://www.technologyreview.com/2026/10/26/1147645/the-focus-on-ai-safety-disasters-ignores-everyday-harms/",
   "external_url": "https://www.technologyreview.com/2026/10/26/1147645/the-focus-on-ai-safety-disasters-ignores-everyday-harms/",
   "title": "The focus on AI safety disasters ignores everyday harms",
   "content_text": "Consider what passes for success in technology policy. An AI company promises that its chatbot will not encourage suicide. A social media platform makes it more difficult for children to buy fentanyl. These are…",
   "language": "en",
   "_wakenex": {
    "source": "mittr-ai",
    "source_name": "MIT Technology Review · AI",
    "group": "coverage",
    "why_en": "Risk attention concentrates on dramatic failures; diffuse everyday harms are structurally undercounted.",
    "why_zh": "风险关注集中于戏剧性事故，日常弥散性伤害在评估中系统性被低估。",
    "title_zh": "对 AI 安全灾难的关注忽视了日常伤害",
    "summary_zh": "想想技术政策领域什么算作成功。一家 AI 公司承诺其聊天机器人不会鼓励自杀。一家社交媒体平台让儿童更难购买芬太尼。这些……",
    "original_language": "en"
   },
   "date_published": "2026-10-26T09:01:00Z"
  },
  {
   "id": "742d30efdc4090c1",
   "url": "https://openai.com/index/openai-cut-incident-investigation",
   "external_url": "https://openai.com/index/openai-cut-incident-investigation",
   "title": "How Ultrafast cut incident investigation from hours to minutes at OpenAI",
   "content_text": "See how OpenAI engineers use Codex Ultrafast to investigate incidents in minutes. Explore faster security analysis, code refactoring, and daily workflows.",
   "language": "en",
   "_wakenex": {
    "source": "openai-news",
    "source_name": "OpenAI News",
    "group": "releases",
    "why_en": "AI-assisted incident triage compresses hours of structured diagnosis into minutes via focused queries.",
    "why_zh": "AI辅助事故排查可将结构化诊断从数小时压缩至数分钟。",
    "title_zh": "Ultrafast如何将OpenAI的事故调查时间从数小时缩短至数分钟",
    "summary_zh": "了解OpenAI工程师如何使用Codex Ultrafast在数分钟内完成事故调查，探索更快速的安全分析、代码重构及日常工作流程。",
    "original_language": "en"
   },
   "date_published": "2026-10-26T07:00:00Z"
  }
 ],
 "_wakenex": {
  "generated_at": "2026-10-27T12:59:19+07:00",
  "note": "Items from public sources that passed WAKENEX's two gates (still useful in a year; the reader can think or do one new thing), linked to the original. The items are not WAKENEX claims; _wakenex.why_en / why_zh is WAKENEX's one-line reading of what each changes."
 }
}