{
 "about": "海外 VC、研究机构和公司画的 AI 安全市场图谱：by 谁画的、d 日期、u 原文、cats 分了哪些类、g 对应本图谱 layers.json 的层（all = 全景）。",
 "maps": [
  {
   "by": "Menlo Ventures",
   "t": "Security for AI: The New Wave of Startups Racing to Secure the AI Stack",
   "d": "2024-02",
   "u": "https://menlovc.com/perspective/security-for-ai-genai-risks-and-the-emerging-startup-landscape/",
   "img": "https://menlovc.com/wp-content/uploads/2024/01/security_for_ai_market_map-012624-scaled.webp",
   "type": "vc",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "c2",
    "e7"
   ],
   "cats": "三大类：治理（AI 资产清单、风险评分）、可观测（日志与滥用审计）、安全（AI 防火墙、红队、PII 脱敏、威胁检测、模型扫描、联邦学习与合成数据）。",
   "note": "页面日期 2024-02-06，署名 Menlo Ventures。"
  },
  {
   "by": "Menlo Ventures",
   "t": "Security for Agents and Agents for Security: The Next Cybersecurity Frontier",
   "d": "2025-07",
   "u": "https://menlovc.com/perspective/security-for-agents-and-agents-for-security-the-next-cybersecurity-frontier/",
   "img": "https://menlovc.com/wp-content/uploads/2025/07/ai_agent_security_thumbnail.jpg",
   "type": "vc",
   "topic": "模型与智能体运行时",
   "g": [
    "a2",
    "e1",
    "e2",
    "c6"
   ],
   "cats": "智能体安全按生命周期分四类：构建（代码与行为测试）、身份与授权、运行时防护、观测与治理；另含红队与评测。"
  },
  {
   "by": "Menlo Ventures",
   "t": "Agents for Security: Rethinking Security for the Age of Autonomy",
   "d": "2025-11",
   "u": "https://menlovc.com/perspective/agents-for-security-and-the-opportunity-to-disrupt-securitys-giants/",
   "img": "https://menlovc.com/wp-content/uploads/2025/11/ai_agents_for_security_thumbnail.jpg",
   "type": "vc",
   "topic": "AI 安全运营",
   "g": [
    "e6",
    "e4",
    "e3"
   ],
   "cats": "用 AI 智能体做安全（AI for security）：SOC 运营、SIEM、渗透测试等被智能体冲击的类别，押「新玩家而非巨头」。",
   "note": "偏论点文，预览图是主题图而非公司名单。"
  },
  {
   "by": "Menlo Ventures",
   "t": "Agents for Security: The Tipping Point for Offensive AI",
   "d": "2026-03",
   "u": "https://menlovc.com/perspective/agents-for-security-the-tipping-point-for-offensive-ai/",
   "img": "https://menlovc.com/wp-content/uploads/2026/03/agents_for_offensive_security_market_map-031826-scaled.png",
   "type": "vc",
   "topic": "AI 写代码与攻防",
   "g": [
    "e4",
    "e2"
   ],
   "cats": "面向企业的攻击性安全（自动渗透测试）＋ 面向 AI 智能体的红队（智能体、LLM 两组）。"
  },
  {
   "by": "Insight Partners",
   "who": "George Mathew、Hunter Korn、Ash Tutika、William Blackwell",
   "t": "The race to secure enterprise AI",
   "d": "2025-07",
   "u": "https://www.insightpartners.com/ideas/securing-ai/",
   "img": "https://www.insightpartners.com/wp-content/uploads/2025/07/blog-hero-1024x576.png",
   "type": "vc",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "c2"
   ],
   "cats": "两个阶段：模型开发安全（红队、模型扫描、AI 软件供应链）与模型运行时安全（提示过滤、AI 威胁检测与响应、DLP）；市场图按这两块排公司。",
   "note": "预览图是文章头图，市场图在正文里。"
  },
  {
   "by": "Insight Partners",
   "who": "George Mathew、Hunter Korn、Ash Tutika、William Blackwell",
   "t": "Securing the autonomous future: Trust, safety, and reliability of agentic AI",
   "d": "2025-10",
   "u": "https://www.insightpartners.com/ideas/securing-agentic-ai/",
   "img": "https://www.insightpartners.com/wp-content/uploads/2025/10/AI-Agents-Security-Market-Map-Blog-Hero-4-scaled.jpg",
   "type": "vc",
   "topic": "模型与智能体运行时",
   "g": [
    "a2",
    "c6",
    "e1",
    "c2"
   ],
   "cats": "AI 智能体安全市场图：智能体身份与访问治理、全栈可观测、上下文感知的网络与 API 监控、针对新型威胁的防护、以数据为中心的隐私控制。"
  },
  {
   "by": "Insight Partners",
   "who": "Thomas Krane、William Blackwell",
   "t": "Identity and Access Management (IAM) in the age of AI Agents",
   "d": "2026-03",
   "u": "https://www.insightpartners.com/ideas/iam-ai-agents/",
   "img": "https://www.insightpartners.com/wp-content/uploads/2026/03/TL-Slides-Agent-IAM-hero-1.jpg",
   "type": "vc",
   "topic": "身份与凭证",
   "g": [
    "a2",
    "a1"
   ],
   "cats": "智能体身份来源与追溯、意图监控、动态授权（零常驻权限 / 即时授权）、端点和云安全、MCP 访问。"
  },
  {
   "by": "MMC Ventures",
   "t": "The Who's Who in Responsible AI",
   "d": "2024-04",
   "u": "https://mmc.vc/research/the-whos-who-in-responsible-ai/",
   "img": "https://mmc.vc/wp-content/uploads/2024/04/AI-Security-market-map-update-scaled.jpg",
   "type": "vc",
   "topic": "治理与合规",
   "g": [
    "e7",
    "e1",
    "e2",
    "c2"
   ],
   "cats": "约 125 家公司分三块：AI 治理、风险与合规（GRC）；AI 质量（监控、评测）；AI 安全与隐私。",
   "note": "英国基金；范围是「负责任 AI」，安全只是其中一块，约占一半融资。"
  },
  {
   "by": "Geodesic Capital",
   "who": "Will Horyn",
   "t": "Cybersecurity in an AI-First World",
   "d": "2024-07",
   "u": "https://geodesiccap.com/insight/cybersecurity-in-an-ai-first-world/",
   "type": "vc",
   "topic": "全景",
   "g": [
    "e1",
    "c2",
    "e7"
   ],
   "cats": "GenAI 网络安全风险与对应工具的市场图，列出各环节的主要玩家。",
   "note": "页面上的图是图片，类目名称未能逐项核对；页面 og 预览图 404，故不放图。"
  },
  {
   "by": "Deutsche Telekom Israel（含 T-Capital、AICC）",
   "t": "Global GenAI Cyber Security Market Map",
   "d": "2024-06",
   "u": "https://www.dtisrael.com/genaisecurity",
   "type": "company",
   "topic": "全景",
   "g": [
    "all"
   ],
   "cats": "140 多家，类别含：AI 安全测试与红队、AI 数据安全与隐私、深伪防御、AI 安全运营、模型与供应链安全、应用与智能体安全、治理与风险、评测与可观测、AI 身份与访问、反欺诈、信任与安全。",
   "note": "电信公司的以色列创新团队做的交互图，持续更新；日期取自 VC 索引（2024-06），页面本身无日期，也无预览图。"
  },
  {
   "by": "Alumni Ventures",
   "t": "AI Is Coming: Meet the Startups Building Cyber Defenses for the Age of AI",
   "d": "2025-04",
   "u": "https://www.av.vc/blog/ai-is-coming-meet-the-startups-building-cyber-defenses-for-the-age-of-ai",
   "type": "vc",
   "topic": "全景",
   "g": [
    "c2",
    "e2",
    "e1",
    "e5",
    "e7"
   ],
   "cats": "AI 安全分层：可见性、治理与控制；部署前测试；运行时防护；数据与隐私等。",
   "note": "Premji Invest 同日转载同一篇（2025-04-14）。预览图是库存照片，故不放图。"
  },
  {
   "by": "G+D Ventures",
   "who": "Ena Uffelmann",
   "t": "AI Security Market Map（66 家）",
   "d": "2026-04",
   "u": "https://www.linkedin.com/posts/ena-uffelmann_gd-ventures-ai-security-market-map-ugcPost-7452295325026807809-4_TB",
   "type": "vc",
   "topic": "全景",
   "g": [
    "all"
   ],
   "cats": "66 家 AI 安全公司；后续计划再出智能体身份、可观测与评测、AI 编排三张。",
   "note": "德国 G+D 旗下 VC；完整图在领英帖子里的链接，帖子本身无预览图，领英页面需登录，只核对了搜索摘要。"
  },
  {
   "by": "Criteria Venture Tech",
   "who": "Aleix Pérez",
   "t": "Offensive Security Market Map: The Rise of Autonomous Hacking",
   "d": "2026-04",
   "u": "https://cyberflow.substack.com/p/offensive-security-market-map-the",
   "img": "https://substackcdn.com/image/fetch/$s_!BgRb!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F84f8abb0-9d49-4393-a714-3a785b99846a_1270x714.png",
   "type": "vc",
   "topic": "AI 写代码与攻防",
   "g": [
    "e4"
   ],
   "cats": "60 多家，分三波：老牌（2015 前）、挑战者（2015–2023）、自主攻击（2024 起）。"
  },
  {
   "by": "Criteria Venture Tech",
   "who": "Aleix Pérez",
   "t": "Auto-Remediation market map: The Next Frontier in AppSec",
   "d": "2025-11",
   "u": "https://cyberflow.substack.com/p/auto-remediation-market-map-the-next",
   "img": "https://substackcdn.com/image/fetch/$s_!rwxK!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F27064d09-eaff-4208-89ee-722db51fe065_1321x745.png",
   "type": "vc",
   "topic": "AI 写代码与攻防",
   "g": [
    "e3"
   ],
   "cats": "AI 自动修复漏洞（Auto-Fixers / AI AppSec 工程师）：老牌、AppSec 挑战者、2023 年起的 LLM 原生三波。"
  },
  {
   "by": "TechOperators",
   "t": "The Cyber × AI Map（Cybersecurity + AI Startup Landscape，403 家）",
   "d": "2026-10",
   "u": "https://www.techoperators.com/landscape",
   "img": "https://techoperators.com/og/og-map-1200x630.png",
   "type": "vc",
   "topic": "全景",
   "g": [
    "all"
   ],
   "cats": "403 家 AI×安全公司按 18 个市场画成一张地图，分六块：SOC、代码与攻击、数据与身份、人与欺诈、暴露与风险、保护 AI。",
   "note": "2025-01 起每季更新，早期版本是领英上的 PDF；此为最新交互版，页面无日期，月份取自 VC 索引。"
  },
  {
   "by": "Applied GAI in Security",
   "who": "Brandon Dixon",
   "t": "Agentic Security Marketmap",
   "d": "2024-12",
   "u": "https://applied-gai-in-security.ghost.io/agentic-security-marketmap/",
   "img": "https://storage.ghost.io/c/08/30/08303bb8-5af4-4264-a256-3d315ab3b698/content/images/size/w1200/2024/12/Screenshot-2024-12-04-at-12.38.10.png",
   "type": "research",
   "topic": "AI 安全运营",
   "g": [
    "e6",
    "e3"
   ],
   "cats": "用智能体做安全的创业公司：事件分诊、代码漏洞分析、安全 Copilot/智能体。",
   "note": "个人博客，作者署名取自 VC 索引，页面不署名。"
  },
  {
   "by": "Cloud Security Alliance",
   "t": "Agentic AI Security Innovator Market Map",
   "d": "2026-06",
   "u": "https://cloudsecurityalliance.org/artifacts/csa-innovator-market-map",
   "img": "https://cloudsecurityalliance.org/rails/active_storage/blobs/redirect/eyJfcmFpbHMiOnsiZGF0YSI6NjQ0NjMsInB1ciI6ImJsb2JfaWQifX0=--00893c1a25fe3ebfe2502c2d4f333319a2d5a59b/CSA%20Innovator%20Market%20Map%20-%20Thumbnail.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "a2",
    "c6",
    "e7",
    "e5",
    "e1"
   ],
   "cats": "身份、治理、可观测、供应链完整性、中立护栏五类（共约 68 家）。",
   "note": "页面写 2026-06-04 发布。"
  },
  {
   "by": "Security Ops Wire",
   "t": "AI Agent Security Companies: A Six-Control-Plane Market Map",
   "d": "2026-09",
   "u": "https://securityopswire.com/ai-agent-security-companies-market-map/",
   "img": "https://securityopswire.com/wp-content/uploads/2026/08/The-AI-Agent-Security-Market-Map_-40-Companies-Building-the-New-Security-Stack.png",
   "type": "research",
   "topic": "模型与智能体运行时",
   "g": [
    "a2",
    "e1",
    "e2",
    "c6",
    "e7"
   ],
   "cats": "六个控制面：发现与盘点、智能体身份与访问、运行时防护、护栏与提示层、治理与策略、红队（40 多家）。",
   "note": "页面日期 2026-09-03；标题与站内旧标题略有不同。"
  },
  {
   "by": "Wavestone",
   "t": "2025 AI security solutions Radar",
   "d": "2025-11",
   "u": "https://www.wavestone.com/en/insight/ai-security-solutions-radar-2025/",
   "img": "https://www.wavestone.com/wp-content/uploads/2025/10/radar-ai-solutions.jpg",
   "type": "research",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "e7",
    "c2"
   ],
   "cats": "94 个方案（2024 版 88 个）；新增「AI 防火墙与响应」类，其余含鲁棒性评测、监控、数据保护、可解释性、过滤、风险管理。",
   "note": "法国咨询公司，每年 10–11 月更新；预览图不是雷达图本身。"
  },
  {
   "by": "OWASP GenAI Security Project",
   "t": "AI Security Solutions Landscape for LLM and GenAI Apps（含智能体、红队两份）",
   "d": "2026-03",
   "u": "https://genai.owasp.org/ai-security-solutions-landscape/",
   "type": "research",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "e7"
   ],
   "cats": "按 AI 软件生命周期（规划、开发、测试、发布、部署、运营、监控、治理）列解决方案，另有智能体和红队专题；附可筛选的厂商目录。",
   "note": "季度更新，2025 Q1 起；Q2 2026 版 2026-03 发布。页面无预览图。"
  },
  {
   "by": "Latio",
   "who": "James Berthoty",
   "t": "2025 Latio AI Security Report",
   "d": "2025-06",
   "u": "https://pulse.latio.tech/p/2025-latio-ai-security-report",
   "img": "https://substackcdn.com/image/fetch/$s_!Vui1!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F20f0d28f-9c79-4a3f-a01f-cfc40614290f_1545x2000.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "c2",
    "e1",
    "e2",
    "e6",
    "e3"
   ],
   "cats": "四个域：终端用户数据管控、AI 安全态势管理、应用运行时防护、用 AI 做安全（SOC、AppSec）。",
   "note": "全文付费，类目取自公开摘要与第三方报道。"
  },
  {
   "by": "Latio",
   "who": "James Berthoty",
   "t": "The 2026 Latio AI Security Market Report",
   "d": "2026-09",
   "u": "https://pulse.latio.tech/p/the-2026-latio-ai-security-market",
   "img": "https://substackcdn.com/image/fetch/$s_!aJfa!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb40189f4-ea65-4693-82a5-21ccd14532a1_1200x1200.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "c1",
    "c2",
    "e1",
    "c6"
   ],
   "cats": "按使用场景分：浏览器、终端应用、云托管智能体构建器、第三方开发工具、自家 AI 使用；今年重心从浏览器与代理转到终端。",
   "note": "全文付费，使用场景分类取自厂商新闻稿的转述。"
  },
  {
   "by": "Software Analyst Cyber Research (SACR)",
   "who": "Francis Odum",
   "t": "Deep Dive Into The Security for AI Ecosystem",
   "d": "2024-08",
   "u": "https://softwareanalyst.substack.com/p/deep-dive-into-the-security-for-ai",
   "img": "https://substackcdn.com/image/fetch/$s_!0tYl!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb1a6c6f3-f93d-439b-a0f2-3ff6903bf45b_1177x665.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "c2",
    "e7"
   ],
   "cats": "AI 安全分为「AI 用于安全」与「安全用于 AI」；风险含对抗攻击、数据投毒、模型输出；按模型生命周期（构建到运行时）列厂商。",
   "note": "作者取自 SACR 索引，页面署名为 SACR。"
  },
  {
   "by": "Software Analyst Cyber Research (SACR)",
   "t": "Data Security Platforms: The New Frontier in Cybersecurity & AI",
   "d": "2024-12",
   "u": "https://softwareanalyst.substack.com/p/data-security-platforms-the-new-frontier",
   "img": "https://substackcdn.com/image/fetch/$s_!c6JN!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2fa2f8a0-1c37-42ab-a7fa-20704002f660_865x562.png",
   "type": "research",
   "topic": "数据与隐私",
   "g": [
    "d1",
    "c2"
   ],
   "cats": "完整数据安全项目的类目：数据发现、分类、访问治理、保护、DLP、加密脱敏、监控响应、备份恢复；并比较数据安全平台厂商。",
   "note": "数据安全赛道的图，AI 是背景，不是只讲 AI。"
  },
  {
   "by": "Software Analyst Cyber Research (SACR)",
   "t": "Securing AI/LLMs in 2025: A Practical Guide To Securing & Deploying AI",
   "d": "2025-03",
   "u": "https://softwareanalyst.substack.com/p/securing-aillms-in-2025-a-practical",
   "img": "https://substackcdn.com/image/fetch/$s_!K9EZ!,w_1200,h_675,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6eb17fea-3e39-4661-8980-56e1c1aad9d4_1427x1050.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "c2",
    "e1",
    "e2",
    "e5",
    "e7"
   ],
   "cats": "保护 AI 分两块：保护员工使用 AI；保护模型与 AI 开发生命周期；并列出厂商产品覆盖对照。"
  },
  {
   "by": "Software Analyst Cyber Research (SACR)",
   "who": "Francis Odum、Rafal Kitab",
   "t": "SACR AI SOC Market Landscape For 2025",
   "d": "2025-08",
   "u": "https://softwareanalyst.io/reports/sacr-ai-soc-market-landscape-for-2025",
   "img": "https://cdn.sanity.io/images/69u797xk/production/d78874049e136d6313ada300f4e4ac207eb087cb-1722x1096.png?rect=0,96,1722,904&w=1200&h=630",
   "type": "research",
   "topic": "AI 安全运营",
   "g": [
    "e6"
   ],
   "cats": "AI SOC 三种集成架构：一体化平台（自带数据）、叠加在现有 SIEM 上的连接层、基于浏览器的分析师工作流模拟；逐家点评。",
   "note": "页面日期 2025-08-04。"
  },
  {
   "by": "CB Insights",
   "t": "The AI security gold rush: who will be snapped up next?",
   "d": "2025-09",
   "u": "https://www.cbinsights.com/research/ai-security-aquisition-matrix/",
   "img": "https://research-assets.cbinsights.com/2025/09/25111246/AI-security-FI.png",
   "type": "research",
   "topic": "全景",
   "g": [
    "e1",
    "e2",
    "e5",
    "a2"
   ],
   "cats": "AI 安全独立创业公司的被收购可能性矩阵；背景是大厂 8 起收购（Check Point 收 Lakera 等）。",
   "note": "CB Insights 完整图需登录。"
  },
  {
   "by": "CB Insights",
   "t": "Anthropic's IPO filing maps the AI safety startup market",
   "d": "2026-10",
   "u": "https://www.cbinsights.com/research/newsletter-october-1-2026/",
   "img": "https://research-assets.cbinsights.com/2026/10/01163119/Anthropic-MiniMarketMap-102026-3.png",
   "type": "research",
   "topic": "其他",
   "g": [
    "c6",
    "e2",
    "e7"
   ],
   "cats": "按四类 AI 风险排公司：拒绝停止（智能体管控）、骗过评测（红队与评测）、出现意外能力（可解释性）、操纵人。",
   "note": "AI 安全（safety）范畴，与企业安全（security）有交叉；图中公司为示例，非穷举。"
  },
  {
   "by": "Panther Labs",
   "t": "Who's Leading AI SOC Automation? A Practitioner's Market Map",
   "d": "2026-04",
   "u": "https://panther.com/blog/ai-soc-automation-market-map",
   "img": "https://framerusercontent.com/images/bECYyaf1j2jroija0ScFzkorrSo.jpg",
   "type": "company",
   "topic": "AI 安全运营",
   "g": [
    "e6"
   ],
   "cats": "AI SOC 三种架构：SIEM 原生 AI 智能体、独立 AI SOC 平台、AI 驱动的 MDR；厂商分平台自带 AI、纯 AI SOC 创业公司、老牌 SIEM 加 AI 三组。",
   "note": "Panther 自己做 SIEM，立场不中立。"
  },
  {
   "by": "Axomap",
   "t": "The Agent Stack: Security & Control Landscape（39 家）",
   "d": "2026-09",
   "u": "https://axomap.com/maps/ai-agent-security-control",
   "img": "https://axomap.com/static/hero-landscape-public.png",
   "type": "company",
   "topic": "模型与智能体运行时",
   "g": [
    "a2",
    "c6",
    "e1",
    "e2"
   ],
   "cats": "五类：工具与 MCP 安全、智能体运行时与执行安全、智能体测试与保障、智能体可观测与取证、智能体身份与授权；只收新的纯做这一层的公司。",
   "note": "独立的图谱网站，AI 辅助整理；页面日期 2026-09-03。"
  }
 ]
}