{
 "generated": "2026-11-01",
 "node": {
  "id": "inst:cot-monitorability-position-paper",
  "type": "Instrument",
  "name": "《思维链可监控性》联合立场声明",
  "slug": "cot-monitorability-position-paper",
  "kind": "research",
  "kind_zh": "论文 / 研究",
  "date": "2025-07",
  "summary": "Tomek Korbak 领衔，OpenAI、Google DeepMind、Anthropic、英国 AI 安全研究院、METR、Apollo Research、Redwood Research 共 41 名研究者联署，预警「在连续隐藏状态里做更多推理、而非生成可读思维链」的架构会消解安全监控的价值。",
  "evidence": [
   {
    "kind": "own",
    "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov",
    "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：跨机构安全共识文件，AI架构安全辩论的核心参照系"
   }
  ],
  "entry": "agent",
  "by": "landscape-structure",
  "at": "2026-11-01T10:26:39Z",
  "landscape": [
   {
    "map": "ai-governance",
    "map_title": "AI 治理与问责",
    "card": "player-cot-monitorability-gov",
    "title": "《思维链可监控性》联合立场声明",
    "layer": "行业自律 与安全机制",
    "layer_id": "layer:ai-governance--6",
    "role": "primary"
   }
  ],
  "landscape_maps": [
   "ai-governance"
  ],
  "landscape_n": 1,
  "degree": 0
 },
 "page": "https://xiaopingfeng.com/graph/e/inst/cot-monitorability-position-paper/",
 "out": [
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及OpenAI",
   "direction": "out",
   "id": "org:openai",
   "name": "OpenAI",
   "role": "participant",
   "note": "联署方；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及Google",
   "direction": "out",
   "id": "org:google",
   "name": "Google",
   "role": "participant",
   "note": "联署方（卡片写作 Google DeepMind）；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及Anthropic",
   "direction": "out",
   "id": "org:anthropic",
   "name": "Anthropic",
   "role": "participant",
   "note": "联署方；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及UK AI Security Institute",
   "direction": "out",
   "id": "org:uk-aisi",
   "name": "UK AI Security Institute（AISI）",
   "role": "participant",
   "note": "联署方（卡片写作 英国AI安全研究院）；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及METR",
   "direction": "out",
   "id": "org:metr",
   "name": "METR",
   "role": "participant",
   "note": "联署方；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及Apollo Research",
   "direction": "out",
   "id": "org:apollo-research",
   "name": "Apollo Research",
   "role": "participant",
   "note": "联署方；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "involves",
   "sentence": "《思维链可监控性》联合立场声明涉及Redwood Research",
   "direction": "out",
   "id": "org:redwood-research",
   "name": "Redwood Research",
   "role": "participant",
   "note": "联署方；牵头人Tomek Korbak（个人，未建组织）",
   "status": "agent",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」的卡片「《思维链可监控性》联合立场声明」：OpenAI、Google DeepMind、Anthropic、英国AI安全研究院、METR、Apollo Research、Redwood Research共41名研究者联署",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  },
  {
   "type": "placed_in",
   "sentence": "《思维链可监控性》联合立场声明出现在 Landscape 的「行业自律与安全机制」",
   "direction": "out",
   "id": "layer:ai-governance--6",
   "name": "行业自律与安全机制",
   "reported": false,
   "status": "derived",
   "evidence": [
    {
     "kind": "own",
     "label": "本站页面",
     "text": "Landscape「AI 治理与问责」第 06 层「行业自律与安全机制」：卡片「《思维链可监控性》联合立场声明」",
     "url": "https://xiaopingfeng.com/landscape/ai-governance/#player-cot-monitorability-gov"
    }
   ]
  }
 ],
 "in": []
}