How mature is each company’s AI governance? We score the AI governance a firm’s public record demonstrates (filings, job postings, board and standards disclosures, firm publications and media), every score traceable to its source.
of firms that claim organizational culture show no demonstrated practice of it, the widest claim-only gap of any factor.
16 of 43 claimants · highest of 10 factors · in the record we hold · Fortune 100
36%
Where the Record Comes From
of demonstrated L3+ practice comes from Media Coverage, the largest single source.
325 of 908 L3+ practice acts · T excluded where invalid · Fortune 100 · n=59
3of 10
Talk to Action
governance factors demonstrated by the typical firm, against 5 claimed.
Median firm: 5 factors claimed, 3 demonstrated · each firm and factor counted once · in the record we hold · Fortune 100 · n=92
17%
Who Actually Staffs It
Of the "AI-governance" roles tagged so far, most are build roles. Only a minority actually govern.
282 of 1,642 tagged role clusters govern · LLM-coded, not yet validated against human labels · whole panel
103to 1
Builders vs. Governors
Provisional
For every AI-governance job posted in 2025, employers posted 103 asking for AI skills.
2025 employer career-site postings · 280,097 AI-skills vs 2,718 AI-governance titles · title-based, so governance work under other titles is missed · LinkUp population
69%
Never Governed
Provisional
Of career sites hiring for AI for five years or more, 69% have never posted an AI-governance job title.
1,652 of 2,400 sites with 25+ AI-skills postings, the first by Oct 2021 · of sites that began AI hiring in 2022–23, 5% posted a governance title within two years · title-based
56%
An L3 Claim Is a Coin Flip
Provisional
A firm that claims L3 practice shows it within two years just over half the time.
421 L3+ claims with two years of follow-up · 56% shown within two years · 38% never followed by any demonstrated practice · right-censored at 2026 Q3
60%
Governance Gap Atlas
Provisional
Insurers leave 60% of their AI-governance claims undemonstrated, against 33% for the Nasdaq-100.
Claims with no demonstrated practice, cohorts of 20+ firms · insurance 60% (n=47) · top-50 banks 38% · top-50 tech 37% · Nasdaq-100 33% · channel coverage differs by cohort
80%
Board Oversight
of firms with board AI disclosures show an operative oversight mechanism; more show a mechanism than make an aspirational statement.
of firms that claim organizational culture show no demonstrated practice of it, the widest claim-only gap of any factor.
16 of 43 claimants · highest of 10 factors · in the record we hold · Fortune 100
36%
Where the Record Comes From
of demonstrated L3+ practice comes from Media Coverage, the largest single source.
325 of 908 L3+ practice acts · T excluded where invalid · Fortune 100 · n=59
3of 10
Talk to Action
governance factors demonstrated by the typical firm, against 5 claimed.
Median firm: 5 factors claimed, 3 demonstrated · each firm and factor counted once · in the record we hold · Fortune 100 · n=92
17%
Who Actually Staffs It
Of the "AI-governance" roles tagged so far, most are build roles. Only a minority actually govern.
282 of 1,642 tagged role clusters govern · LLM-coded, not yet validated against human labels · whole panel
103to 1
Builders vs. Governors
Provisional
For every AI-governance job posted in 2025, employers posted 103 asking for AI skills.
2025 employer career-site postings · 280,097 AI-skills vs 2,718 AI-governance titles · title-based, so governance work under other titles is missed · LinkUp population
69%
Never Governed
Provisional
Of career sites hiring for AI for five years or more, 69% have never posted an AI-governance job title.
1,652 of 2,400 sites with 25+ AI-skills postings, the first by Oct 2021 · of sites that began AI hiring in 2022–23, 5% posted a governance title within two years · title-based
56%
An L3 Claim Is a Coin Flip
Provisional
A firm that claims L3 practice shows it within two years just over half the time.
421 L3+ claims with two years of follow-up · 56% shown within two years · 38% never followed by any demonstrated practice · right-censored at 2026 Q3
60%
Governance Gap Atlas
Provisional
Insurers leave 60% of their AI-governance claims undemonstrated, against 33% for the Nasdaq-100.
Claims with no demonstrated practice, cohorts of 20+ firms · insurance 60% (n=47) · top-50 banks 38% · top-50 tech 37% · Nasdaq-100 33% · channel coverage differs by cohort
§ · Sectors — click a band to hide, a tab to restore
Showing 91 of 91
Fortune 100 Publication Status
12/12 sector cohorts hardened. This board is live as a provisional cross-cohort ranking pending final publication signoff.
The Governance Score is the evidence ladder: the mean of P(>=L0)...P(>=L5) across 10 factors, with P(>=k) = 1 - 0.2^N for N distinct sources at >= k.
Rungs above a factor’s highest source are exactly 0; a lone top-rung source gets half credit.
Coverage (factors seen / 10) is shown on hover.
The floor tick marks the independently corroborated score — [floor, DA] is corroboration depth, not a confidence interval.
It is demonstrated governance — it correlates with disclosure, so a mature non-discloser scores low.
Cross-sector ranks are provisional (llm_inferred), pending the invariance + coverage checks; L5 (Ethical Vanguard *) is defined within a firm’s own sector.