Anthropic

{
  "id": "anthropic",
  "wikiId": "E22",
  "path": "/knowledge-base/organizations/anthropic/",
  "filePath": "knowledge-base/organizations/anthropic.mdx",
  "title": "Anthropic",
  "quality": 74,
  "readerImportance": 52.3,
  "researchImportance": 57.5,
  "tacticalValue": 92,
  "contentFormat": "article",
  "causalLevel": null,
  "lastUpdated": "2026-06-17",
  "dateCreated": "2026-02-15",
  "summary": "Comprehensive reference page on Anthropic covering financials (\\$380B valuation, \\$14B ARR at Series G growing to \\$19B by March 2026), safety research (Constitutional AI, mechanistic interpretability, model welfare), governance (LTBT structure), controversies (alignment faking at 12%, RSP rollback), and competitive positioning (42% enterprise coding share). Highly concrete with specific numbers throughout but primarily descriptive compilation rather than original analysis.",
  "description": "An AI safety company founded by former OpenAI researchers that develops frontier AI models while pursuing safety research, including the Claude model family, Constitutional AI, and mechanistic interpretability.",
  "ratings": {
    "focus": 7.5,
    "novelty": 4.2,
    "rigor": 6.8,
    "completeness": 7.8,
    "concreteness": 8.1,
    "actionability": 4.5,
    "objectivity": 6.2
  },
  "category": "organizations",
  "subcategory": "labs",
  "clusters": [
    "ai-safety",
    "community",
    "governance"
  ],
  "metrics": {
    "wordCount": 5416,
    "tableCount": 4,
    "diagramCount": 0,
    "internalLinks": 87,
    "externalLinks": 15,
    "footnoteCount": 48,
    "bulletRatio": 0.02,
    "sectionCount": 42,
    "hasOverview": true,
    "structuralScore": 14
  },
  "suggestedQuality": 93,
  "updateFrequency": 3,
  "evergreen": true,
  "wordCount": 5416,
  "unconvertedLinks": [
    {
      "text": "Harvard Law",
      "url": "https://corpgov.law.harvard.edu/2023/10/28/anthropic-long-term-benefit-trust/",
      "resourceId": "357cf00ad44eea37",
      "resourceTitle": "Anthropic Long-Term Benefit Trust"
    },
    {
      "text": "Anthropic",
      "url": "https://www.anthropic.com/news/claude-opus-4-5",
      "resourceId": "57f01cae307e1cb1",
      "resourceTitle": "Introducing Claude Opus 4.5"
    },
    {
      "text": "TechCrunch",
      "url": "https://techcrunch.com/2025/07/31/enterprises-prefer-anthropics-ai-models-over-anyone-elses-including-openais/",
      "resourceId": "3a07423e8bf204c2",
      "resourceTitle": "Enterprises prefer Anthropic's AI models over anyone else's, including OpenAI's"
    },
    {
      "text": "Anthropic",
      "url": "https://transformer-circuits.pub/2024/scaling-monosemanticity/index.html",
      "resourceId": "426fcdeae8e2b749",
      "resourceTitle": "Anthropic's dictionary learning work"
    },
    {
      "text": "MIT TR",
      "url": "https://www.technologyreview.com/2026/01/12/1130003/mechanistic-interpretability-ai-research-models-2026-breakthrough-technologies/",
      "resourceId": "3a4cf664bf7b27a8",
      "resourceTitle": "Mechanistic interpretability: 10 Breakthrough Technologies 2026 | MIT Technology Review"
    },
    {
      "text": "Bank Info Security",
      "url": "https://www.bankinfosecurity.com/models-strategically-lie-finds-anthropic-study-a-27136",
      "resourceId": "de18440757f72c95",
      "resourceTitle": "Models Can Strategically Lie, Finds Anthropic Study"
    },
    {
      "text": "Axios",
      "url": "https://www.axios.com/2025/05/23/anthropic-ai-deception-risk",
      "resourceId": "e76f688da38ef0fd",
      "resourceTitle": "Axios: Anthropic AI Deception Risk (May 2025)"
    },
    {
      "text": "Glassdoor",
      "url": "https://www.glassdoor.com/Overview/Working-at-Anthropic-EI_IE8109027.11,20.htm",
      "resourceId": "150ce5e988bc1e00",
      "resourceTitle": "Glassdoor: Working at Anthropic"
    },
    {
      "text": "Glassdoor",
      "url": "https://www.glassdoor.com/Overview/Working-at-Anthropic-EI_IE8109027.11,20.htm",
      "resourceId": "150ce5e988bc1e00",
      "resourceTitle": "Glassdoor: Working at Anthropic"
    },
    {
      "text": "Glassdoor",
      "url": "https://www.glassdoor.com/Overview/Working-at-Anthropic-EI_IE8109027.11,20.htm",
      "resourceId": "150ce5e988bc1e00",
      "resourceTitle": "Glassdoor: Working at Anthropic"
    },
    {
      "text": "Glassdoor",
      "url": "https://www.glassdoor.com/Overview/Working-at-Anthropic-EI_IE8109027.11,20.htm",
      "resourceId": "150ce5e988bc1e00",
      "resourceTitle": "Glassdoor: Working at Anthropic"
    },
    {
      "text": "Glassdoor",
      "url": "https://www.glassdoor.com/Overview/Working-at-Anthropic-EI_IE8109027.11,20.htm",
      "resourceId": "150ce5e988bc1e00",
      "resourceTitle": "Glassdoor: Working at Anthropic"
    }
  ],
  "unconvertedLinkCount": 12,
  "convertedLinkCount": 0,
  "backlinkCount": 293,
  "citationHealth": {
    "total": 42,
    "withQuotes": 30,
    "verified": 29,
    "accuracyChecked": 29,
    "accurate": 11,
    "inaccurate": 6,
    "avgScore": 0.9623388965924581
  },
  "hallucinationRisk": {
    "level": "medium",
    "score": 45,
    "factors": [
      "biographical-claims",
      "well-cited"
    ]
  },
  "entityType": "organization",
  "redundancy": {
    "maxSimilarity": 20,
    "similarPages": [
      {
        "id": "anthropic-investors",
        "title": "Anthropic (Funder)",
        "path": "/knowledge-base/organizations/anthropic-investors/",
        "similarity": 20
      },
      {
        "id": "labs-overview",
        "title": "Frontier AI Labs (Overview)",
        "path": "/knowledge-base/organizations/labs-overview/",
        "similarity": 20
      },
      {
        "id": "deep-learning-era",
        "title": "Deep Learning Revolution (2012-2020)",
        "path": "/knowledge-base/history/deep-learning-era/",
        "similarity": 19
      },
      {
        "id": "openai",
        "title": "OpenAI",
        "path": "/knowledge-base/organizations/openai/",
        "similarity": 19
      },
      {
        "id": "agentic-ai",
        "title": "Agentic AI",
        "path": "/knowledge-base/capabilities/agentic-ai/",
        "similarity": 18
      }
    ]
  },
  "changeHistory": [
    {
      "date": "2026-02-26",
      "branch": "claude/claims-driven-improvements",
      "title": "Auto-improve (standard): Anthropic",
      "summary": "Improved \"Anthropic\" via standard pipeline (570.8s). Quality score: 74. Issues resolved: Section duplication: 'Competitive Positioning' subsection un; Section duplication: 'Safety Levels' subsection repeats cont; Section duplication: The 'Quick Financial Context' subsectio.",
      "duration": "570.8s",
      "cost": "$5-8"
    },
    {
      "date": "2026-02-24",
      "branch": "feat/stale-fact-detection-581-582",
      "title": "Batch content fixes + stale-facts validator + 2 new validation rules",
      "summary": "(fill in)",
      "pr": 924,
      "model": "claude-sonnet-4-6"
    },
    {
      "date": "2026-02-19",
      "branch": "claude/resolve-issue-203-d8IBd",
      "title": "Calc pipeline iteration: fix range facts, index mismatch, prompt quality",
      "summary": "Ran `crux facts calc` on anthropic-valuation and anthropic pages post-implementation, discovered and fixed three bugs: (1) range-valued facts ({min: N}) invisible to LLM and evaluator, (2) proposal-to-pattern index mismatch causing wrong validation expected values, (3) over-wide originalText proposals including JSX tags or prose. Applied validated Calc replacements to two pages (openai.39d6868e/$500B valuation now computes correctly).",
      "model": "sonnet-4",
      "duration": "~40min"
    },
    {
      "date": "2026-02-18",
      "branch": "claude/source-unsourced-facts-RecGw",
      "title": "Source unsourced facts",
      "summary": "Sourced 25 of 30 previously unsourced facts across all 4 fact files (anthropic, sam-altman, openai, jaan-tallinn). Created 21 new resource entries in news-media.yaml and ai-labs.yaml with proper SHA256-based IDs. Added 8 new publications (Bloomberg, The Information, Quartz, Benzinga, Britannica, World, Sherwood News). Fixed date accuracy issues (Worldcoin stats from 2024 to 2025-05, OpenAI revenue from Oct to Jun 2024) and improved notes. Source coverage improved from 29% to 88%.",
      "model": "opus-4-6",
      "duration": "~45min"
    },
    {
      "date": "2026-02-18",
      "branch": "claude/review-pr-216-P4Fcu",
      "title": "Fix audit report findings from PR #216",
      "summary": "Reviewed PR #216 (comprehensive wiki audit report) and implemented fixes for the major issues it identified: fixed 181 path-style EntityLink IDs across 33 files, converted 164 broken EntityLinks (referencing non-existent entities) to plain text across 38 files, fixed a temporal inconsistency in anthropic.mdx, and added missing description fields to 53 ai-transition-model pages."
    },
    {
      "date": "2026-02-18",
      "branch": "claude/highlight-stakeholder-table-VtY0t",
      "title": "Create dedicated Anthropic stakeholder page",
      "summary": "Created a new dedicated `anthropic-stakeholders` page with the most shareable ownership tables (all stakeholders with stakes, values, EA alignment), added a condensed stakeholder summary to the top of the main Anthropic page, and wrote 4 proposed GitHub issues for broader system changes (datasets infrastructure, importance metrics rethink, concrete data expansion, continuous maintenance).",
      "model": "opus-4-6",
      "duration": "~30min"
    },
    {
      "date": "2026-02-18",
      "branch": "claude/fact-hash-ids-UETLf",
      "title": "Migrate fact IDs from human-readable to hash-based",
      "summary": "Migrated all canonical fact IDs from human-readable slugs (e.g., `revenue-arr-2025`) to 8-char random hex hashes (e.g., `55d88868`), matching the pattern used by resources. Updated all YAML files, MDX references, build scripts, tests, LLM prompts, and documentation.",
      "model": "opus-4-6",
      "duration": "~45min"
    },
    {
      "date": "2026-02-18",
      "branch": "claude/audit-webpage-errors-11sSF",
      "title": "Fix factual errors found in wiki audit",
      "summary": "Systematically audited ~35+ high-risk wiki pages for factual errors and hallucinations using parallel background agents plus direct reading. Fixed 13 confirmed errors across 11 files."
    }
  ],
  "coverage": {
    "passing": 8,
    "total": 13,
    "targets": {
      "tables": 22,
      "diagrams": 2,
      "internalLinks": 43,
      "externalLinks": 27,
      "footnotes": 16,
      "references": 16
    },
    "actuals": {
      "tables": 4,
      "diagrams": 0,
      "internalLinks": 87,
      "externalLinks": 15,
      "footnotes": 48,
      "references": 102,
      "quotesWithQuotes": 30,
      "quotesTotal": 42,
      "accuracyChecked": 29,
      "accuracyTotal": 42
    },
    "items": {
      "summary": "green",
      "schedule": "green",
      "entity": "green",
      "editHistory": "green",
      "overview": "green",
      "tables": "amber",
      "diagrams": "red",
      "internalLinks": "green",
      "externalLinks": "amber",
      "footnotes": "green",
      "references": "green",
      "quotes": "amber",
      "accuracy": "amber"
    },
    "editHistoryCount": 8,
    "ratingsString": "N:4.2 R:6.8 A:4.5 C:7.8"
  },
  "readerRank": 281,
  "researchRank": 235,
  "recommendedScore": 183.03
}

{
  "wikipedia": "https://en.wikipedia.org/wiki/Anthropic",
  "lesswrong": "https://www.lesswrong.com/tag/anthropic-org",
  "wikidata": "https://www.wikidata.org/wiki/Q116758847",
  "grokipedia": "https://grokipedia.com/page/Anthropic"
}