CAINCCorporate AI Narrative CorpusPILOT / 0.8
RELEASE 0.8 · 4 OCTOBER 2026

Corpus methodology

Rights-aware longitudinal research across corporate narrative, governance, public engagement, and statutory disclosure.

Current release

165 corporate document records cover 2017–2026. The local aggregate layer contains 1,303,196 analyzed words; the public searchable layer contains 58,757 words.

Access classes

  1. Public searchable text: complete text with a documented reuse basis.
  2. Local aggregate analysis: source prose stays in the research workspace; the site publishes metadata, totals, controlled-vocabulary counts, and original links.
  3. Metadata only: provenance and limited quotations where appropriate; no document-level frequency denominator.

Accountability tiers

  1. Promotional and public narrative.
  2. Voluntary self-governance.
  3. Statements to public institutions.
  4. Statutory or legally structured disclosure.

Current analytical coverage

The local layer includes Item 1A Risk Factors from 70 SEC annual filings for Microsoft, NVIDIA, AMD, Alphabet, Amazon, Meta, and Tesla, plus 29 corporate governance and safety texts. 23 Tier 2 publications are dated 2017–2024, with at least two in every year. Frequency equals matches divided by analyzed words multiplied by 10,000. A missing year remains “No data,” not zero.

Governance vocabulary

The published concept measure sums disclosed exact matches for governance, regulation, compliance, responsibility, accountability, transparency, ethics, fairness, oversight, audit, policy, principle, and standard word forms. Individual exact-word queries remain available.

Rights

CAINC-authored metadata, taxonomies, and annotations are released under CC BY 4.0. Source documents retain their original copyrights and terms.

Open the full interactive methodology