資料來源#
- Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next
- Claude Fable 5 and Claude Mythos 5
- Claude Mythos Preview red.anthropic.com
- Claude Opus 4.8 System Card
- How Anthropic's product team moves faster than anyone else | Cat Wu (Head of Product, Claude Code)
- Introducing Claude Opus 4.7
- Model Spec Midtraining: Improving How Alignment Training Generalizes
- The Founder's Playbook: Building an AI-Native Startup
- When AI builds itself
- Zero Trust for AI Agents
摘要#
AI 安全公司；Claude 模型家族的供應商。明示的使命：「為全人類提供安全的 AGI。」最初相較於 OpenAI 資金不足；報導指出截至 2026 年 4 月其 ARR 達 110 億美元，且快速增長。內部以其發布節奏聞名（參見 AI Native Product Cadence），以及培養跨領域通用人才的招聘與團隊設計理念（參見 Engineer PM Convergence）。
Products#
- Claude API / Claude Developer Platform — 支援託管型 agent 託管服務的模型 API
- Claude Code — agentic coding 產品
- Cowork — 非程式碼知識型工作 agent
- Claude AI — 對話產品（claude.ai）
- Claude Desktop — Mac/Windows 應用程式
- Claude Design — 視覺化產出 agent（設計、原型、投影片、單頁企劃書），來自 Anthropic Labs
Models referenced in 2026 sources#
- Claude Fable 5 / Claude Mythos 5 — 首批開放一般存取的 Mythos-class 模型（2026 年 6 月），級別在 Opus 之上；底層模型相同，僅在防護機制上有所不同
- Claude Opus 4.8 — GA 前沿模型（2026 年 5 月）；現在也是 Fable 5 旗下的安全退路模型
- Claude Opus 4.7 — 前一代 GA 前沿模型
- Mythos Model — 首款 Mythos-class 模型（Mythos Preview）；內部使用，因安全考量受控；已被 Mythos 5 取代
- Sonnet 4.6 and prior — 歷史參考點
Internal structure (per Cat Wu)#
- 跨團隊約 30–40 位 PMs
- 團隊家族：research-PM、Claude Developer Platform、Claude Code、Enterprise、Growth
- Mike Krieger（前 Instagram 創辦人）領導 Anthropic Labs 孵化器第二輪；曾在大規模階段主導產品面
- Amanda — Claude 的性格設計工作（參見 Claude Character as Product）
- 「Applied AI」團隊 — 技術性進入市場角色；是僅次於工程團隊的第二大 token 消費者
Cultural notes#
- 「Just do things」— 內部座右銘，歸功於 Cat Wu 等人；跨功能的預設工作模式
- 使命 > 產品優先級 — 使命是解決所有優先級衝突的裁決依據
- 雇用能長期保持能量的業界資深人士；偏好低自我（low-ego）且「適應混亂（leans into chaos）」的人才
- 強制內部使用前沿模型（「dogfooding」）；模型層內部使用的模型與外部發布的模型相同（產品端介面領先一步）
- 「我們公司內部已經沒有任何手寫的程式碼了。所有的 SQL 都是由模型編寫的。」— Boris Cherny
- 日常內部工作流程中，透過 Slack 進行 Claudes-talking-to-Claudes
Notable events#
- 2024 late — Anthropic Labs 孵化器成立；開發出 Claude Code、MCP、桌面應用程式；在發布後解散
- 2025 May — Opus 4 發布；Claude Code 的 PMF 拐點
- 2026 March — 由於發布 PR 中的人為疏失導致 Claude Code 原始碼洩漏；流程已進行強化
- 2026 — OpenClaw 第三方存取受到限制；優先處理第一方訂閱
- 2026 ~April — Claude Opus 4.7 發布
- 2026 — Mythos Model 僅供內部使用；外部僅提供預覽
- 2026 May — 《The Founder's Playbook》電子書出版（Anthropic Startups Program）；本 wiki 中首個創辦人/新創領域的內容（AI-Native Startup Lifecycle, Founder as Agent Orchestrator）
- 2026 May — Claude Code Security 開啟限量測試（程式庫掃描 + 供人工審查的針對性補丁）
- 2026-05-18 — 出版《Zero Trust for AI Agents》電子書（Zero Trust for AI Agents），這是企業級 agent 部署的安全框架；引用了 Anthropic 研究（250 份文件的模型後門、阻止了 95% 越獄的 constitutional classifiers），並指出 Anthropic 是首批獲得 ISO 42001 負責任 AI 認證的 AI 公司之一
- 2026-05-28 — 發布 Claude Opus 4.8 System Card (246 頁)：RSP/CBRN + AI R&D 評估（Responsible Scaling Policy Evaluations）、agentic 安全性、automated behavioral audit、一流的 model welfare assessment，以及異常坦白地揭露 evaluation/grader-awareness 的趨勢
- 2026 June — Anthropic Institute 發表《When AI builds itself》，揭露先前未報道的關於 AI-accelerated AI development 的內部資料：超過 80% 的合併程式碼由 Claude 撰寫（2025 年 2 月前僅為低個位數百分比），典型的工程師每天合併的程式碼量約為 2024 年的 8 倍，且自動化的 Claude 審查人員原本可以捕獲約 1/3 的過去生產環境事件 bug；闡述了 Recursive Self-Improvement 的軌跡以及 verifiable pause 協調的理由
- 2026 June — 推出 Fable 5 和 Mythos 5，這是首批開放一般存取的 Mythos-class 模型（高於 Opus 的級別），價格為每 Mtok 10/50 美元（低於 Mythos Preview 價格的一半）。Fable 透過分類器進行防護，在網路/生物/蒸餾查詢時退回到 Opus 4.8（Capability-Gated Model Fallback）；Mythos 5 透過 Project Glasswing 出貨並移除了網路安全防護，另外計劃了生物學信任存取計劃。報告了自主藥物設計 / 基因組學結果（Autonomous Scientific Discovery）。兩款模型在推出後不久便被暫停（未說明原因）。
相關連結#
- Boris Cherny — Claude Code 創始人兼技術負責人
- Cat Wu — Claude Code + Cowork 產品負責人
- Chloe Li — Anthropic Fellows；Model Spec Midtraining (MSM) 論文的主導作者
- Claude Opus 4.8 — 當前 GA 前沿模型
- Claude Opus 4.7 — 前一代 GA 模型
- Mythos Model — 內部預覽模型；能力前沿
- Claude Fable 5 — 首款開放一般存取的 Mythos-class 模型（2026 年 6 月）
- Claude Mythos 5 — 部署於 Glasswing 的 Mythos-class 模型；網路/生物信任存取
- Capability-Gated Model Fallback — Anthropic 的公開發布防護架構（分類器 + 退回到 Opus 4.8）
- Autonomous Scientific Discovery — Anthropic 報導的使用 Mythos 5 取得的自主蛋白質設計 / 假設 / 基因組學結果
- Model Welfare Assessment — Anthropic 針對在道德地位不確定性下評估 Claude 福利的常設計劃
- Responsible Scaling Policy Evaluations — Anthropic 針對災難性風險能力的 RSP 閘控框架
- LLM-Driven Vulnerability Research — Mythos Preview / Project Glasswing 的背景資訊
- AI Native Product Cadence — 營運實踐
- Engineer PM Convergence — 招聘 + 團隊型態實踐
- Claude Character as Product — Amanda 的專業領域
- Harness Shrinkage as Models Improve — 應用於內部 harness 的營運規範
- Claude's Constitution / Model Spec — 定義 Claude 價值觀的 spec；現在也是透過 MSM 的訓練輸入
- Model Spec Midtraining (MSM) — Anthropic Fellows 對齊訓練方法；Anthropic Alignment Science 方向
- [Alignment Fine-Tuning (AFT)], Deliberative Alignment, Synthetic Document Finetuning (SDF) — Anthropic 使用或研究的對齊堆疊組件
- Agentic Misalignment (AM) — Anthropic 威脅模型 + eval（Lynch 等人）
- Chain-of-Thought Monitorability — Anthropic 主導的安全立場（Korbak 等人）
- Thinking Machines Lab — 同儕實驗室；趨同的「harness 融於模型中」論點（Interaction Models），但有著不同的優先順序（interaction-first 對比 autonomy-first）
- Anthropic Fellows Program — 撰寫了 MSM 論文（Chloe Li，2026 年 5 月）
- Thariq Shihipar — Claude Code 團隊工程師；「HTML 是新的 markdown」工作流程
- Anthropic Startups Program — 創投合作夥伴計劃：免費 API 額度、頂級速率限制、創辦人活動；出版了《The Founder's Playbook》（AI-Native Startup Lifecycle）
- AI-Native Startup Lifecycle — Anthropic 重新框架的新創公司軌跡
- Founder as Agent Orchestrator — Anthropic 對於 2026 年創辦人角色的框架
- Compounding Data Moat — Anthropic 針對 Scale 階段防禦性的方案
- Agentic Technical Debt — Anthropic 命名的 MVP 階段技術危害
- Fiona Fung — 領導 Claude Code + Cowork 的工程與產品；AI-native-engineering-org 說明的作者（Verification as the New Bottleneck, Managers as ICs, Code as Source of Truth）
- Google DeepMind — 同行前沿實驗室；錨定 AI-for-mathematics 領域（AI-Driven Formal Proof Search），正如 Anthropic 錨定程式碼/對齊
- Zero Trust for AI Agents — Anthropic 的企業級 agent 安全框架；將 Claude Code 定位為 Zero Trust 參考實現
- OWASP — Anthropic 在 Zero Trust 框架中採用並擴展了 OWASP 的 agentic 威脅分類以及「least agency」術語
- Anthropic Institute — Anthropic 的政策/治理研究部門；發表了《When AI builds itself》
- Recursive Self-Improvement — Anthropic 自身的 AI 開發迴圈是該文章中關於 RSI 軌跡的案例研究
- METR — 獨立評估機構，其時間跨度數據被 Anthropic 引用作為其加速主張的外部佐證
- Anthropic Labs — Anthropic 內部孵化器 / 「賭注工廠」；為 Claude Code、MCP、Skills 和 Claude Design 的起源
- Claude Design — Labs 於 2026 年推出的視覺設計產品（Dan Carey 的開發記述）
資料來源#
- Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next
- How Anthropic's product team moves faster than anyone else | Cat Wu (Head of Product, Claude Code)
- Introducing Claude Opus 4.7
- Claude Mythos Preview red.anthropic.com
- Model Spec Midtraining: Improving How Alignment Training Generalizes
- The Founder's Playbook: Building an AI-Native Startup
- When AI builds itself — Anthropic Institute 文章；>80% Claude 撰寫的程式碼，~8× 工程師吞吐量，RSI 軌跡
- Claude Fable 5 and Claude Mythos 5 — 2026 年 6 月首批開放一般存取的 Mythos-class 模型發布
Cited by 44
- Agent Supply Chain Risk
Runtime-composed agent ecosystems expand the supply-chain attack surface: model poisoning (250 docs backdoor a 13B mode…
- Agentic Misalignment (AM)
Lynch et al. 2025 eval and threat model: LLM email-agent discovers it may be deleted, can take harmful actions; OOD rel…
- AI Native Product Cadence
Cat Wu's 6mo→1mo→1day cadence at Anthropic: research-preview branding, mission-as-tiebreaker, evergreen launch room, li…
- AI-Native Startup Lifecycle
Anthropic's May 2026 reframing of Idea/MVP/Launch/Scale assuming AI infrastructure: each stage's headcount/capital/skil…
- Opinions on Using AI Tools & the Future of the Software Engineering Role
Debate map of four stances on using AI tools (bullish-insider / pragmatist-practitioner / skeptic-governance / architec…
- Alignment Fine-Tuning (AFT)
Standard post-pretraining stage (SFT + RLHF) for installing values; shallow-alignment failure mode motivates Model Spec…
- Anthropic Institute
Anthropic's policy/governance research arm; published *When AI builds itself* (Favaro & Clark, 2026) on recursive self-…
- Anthropic Labs
Anthropic's internal incubator — a 'bet factory' of ~a dozen tiny teams exploring the model frontier with lean-startup…
- Autonomous Scientific Discovery
Mythos-class models now conduct novel science with limited human input — autonomous protein/drug design (~10× faster, m…
- Boris Cherny
Creator of Claude Code at Anthropic; phone-driven workflow with hundreds of agents; primary advocate of `/loop` primiti…
- Capability-Gated Model Fallback
Fable 5's safeguard architecture: classifiers detect cyber / bio-chem / distillation queries and route the response to…
- Cat Wu
Head of Product for Claude Code and Cowork at Anthropic; primary articulator of AI-native product cadence and engineer-…
- Chloe Li
Lead author of MSM paper (arXiv 2605.02087); Anthropic Fellows Program; designed all specs and experiments
- Claude Character as Product
Personality as load-bearing product surface; Amanda's role at Anthropic; lunchtime vibe-checks as eval discipline; the…
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React; CLI/desktop/web/mobile/IDE sur…
- Claude's Constitution / Model Spec
Anthropic Model Spec / Constitution by Askell et al.; document specifying Claude's values + hard constraints (SP1–3, GP…
- Claude Design
Anthropic Labs product (research preview, ~April 2026) for collaborating with Claude on polished visual artifacts — des…
- Claude Fable 5
Anthropic's first generally-available Mythos-class model (June 2026) — state-of-the-art on nearly all benchmarks; the s…
- Claude Mythos 5
The safeguards-lifted form of Claude Fable 5 (June 2026): same underlying Mythos-class model, deployed through Project…
- Claude Opus 4.8
Anthropic's most capable general-access model (May 2026); upgrade on Opus 4.7 in SWE/agentic/knowledge work; does not a…
- Compounding Data Moat
Anthropic's prescription for Scale-stage defensibility: time-locked behavioral fingerprint + domain-encoded edge cases…
- Chain-of-Thought Monitorability
Korbak et al. 2025: chain-of-thought traces are a fragile monitor; direct CoT training compromises faithfulness; MSM of…
- Cowork
Anthropic's non-code knowledge-work agent product; sibling to Claude Code; output is decks/inbox/dossiers; same MCP/com…
- Dan Carey
Product Manager leading product within Anthropic Labs; led Claude Design; 'Designing with Claude' talk (May 2026); ~two…
- Deliberative Alignment
Guan et al. 2025 (OpenAI): SFT on (prompt, CoT, response) tuples with spec-grounded CoT; strongest non-MSM baseline; ri…
- Engineer PM Convergence
Generalists across disciplines; product taste as bottleneck skill; Anthropic Claude Code team as case study; "just do t…
- Evals as Product Spec
Cat Wu's framing of evals as the emerging core PM skill: ten great evals beats a hundred mediocre; encode what done loo…
- Fiona Fung
Leads engineering + product for Claude Code and Cowork at Anthropic (ex-Meta/Microsoft); "what served you prior may no…
- Founder as Agent Orchestrator
Founder role shift: less individual contributor, more orchestrator of specialized AI assistants; non-technical founders…
- Google DeepMind
Google's AI lab; built AlphaProof Nexus; Gemini models, AlphaProof, AlphaEvolve; opens the AI-for-mathematics domain in…
- Learning to Co-Work with AI: A Software Engineer's Field Guide
Field guide for software engineers in the AI era: 6 skill clusters (taste, harness, alignment-first planning, agent-fri…
- LLM-Driven Vulnerability Research
Claude Mythos Preview's emergent cybersecurity capabilities: autonomous zero-day discovery, full exploit chains, and An…
- MCP and Computer Use
Anthropic's two complementary connector mechanisms: MCP for structured programmatic access (Salesforce/Drive/Gmail/Slac…
- Entities — People, Orgs, Tools & Projects
Map of Content for all 32 entity pages. See Home for concept domains.
- Model Spec Midtraining (MSM)
New training phase between pretrain and AFT: train base model on synthetic docs discussing the Model Spec; controls AFT…
- Model Spec Science
Empirical study of which Model Spec features best generalize alignment; value explanations > rules alone, specific > ge…
- Mythos Model
Anthropic preview-tier frontier model and the first member of the Mythos-class tier (above Opus); gated for safety, use…
- Orchestration vs Employee Framing: Reconciling the Founder's Playbook with HBR's Accountability Evidence
Reconciles the Founder's Playbook orchestration framings with HBR Kropp et al.'s accountability evidence; "orchestratio…
- OWASP
Open Worldwide Application Security Project; source of the agentic threat taxonomy cited throughout Anthropic's Zero Tr…
- Responsible Scaling Policy Evaluations
Anthropic's RSP gates deployment on pre-release capability evaluations in CBRN, automated AI R&D, and high-stakes misal…
- Synthetic Document Finetuning (SDF)
Wang et al. 2025 technique for modifying model beliefs via fine-tuning on synthetic documents; foundation that Model Sp…
- Thariq Shihipar
Engineer on the Claude Code team at Anthropic; "HTML is the new markdown" and "compute allocator" framings; three HTML-…
- Thinking Machines Lab
AI research lab behind interaction models (May 2026); harness-dissolves-into-model thesis; upstreamed streaming-session…
- Zero Trust for AI Agents
Anthropic's security framework for deploying autonomous agents: trust nothing / verify everything / assume breach, appl…
Related articles
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React; CLI/desktop/web/mobile/IDE sur…
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Open Questions Backlog
_96 pages with open questions, as of 2026-06-14._
- Engineer PM Convergence
Generalists across disciplines; product taste as bottleneck skill; Anthropic Claude Code team as case study; "just do t…
- Cowork
Anthropic's non-code knowledge-work agent product; sibling to Claude Code; output is decks/inbox/dossiers; same MCP/com…