Only Hugging Face and Stripe reach the AA tier in ENTRA's August 2026 pay transparency audit of 20 major AI employers. The remaining 18 companies — including some of the highest-valued and most publicly scrutinized names in artificial intelligence — publish compensation information selectively, inconsistently, or only where state and national law compels them to. Twelve of the 20 companies score B or BB, meaning proactive disclosure on any of the five scored dimensions is largely absent. The gap between Hugging Face's score of 85 and Modal's score of 30 spans 55 points on a 100-point scale. In a sector that competes more aggressively for talent than any other and that routinely positions culture and mission as differentiators, that gap reflects a structural contradiction: AI employers ask candidates to make consequential career decisions without the information those candidates need to make them well.
The Scoring Methodology
This index scores each company across five dimensions that together capture what meaningful pay transparency looks like — not merely what the law requires.
Salary Band Disclosure (30 points) measures whether companies publish salary ranges on job postings globally, not only in the US states that mandate it. A company posting ranges only in Colorado and New York receives substantially fewer points than one that publishes ranges on every global posting without being required to.
Equity Compensation Transparency (25 points) measures whether companies publicly disclose equity grant ranges, vesting schedules, and dilution policies. For private companies, this includes whether equity terms are explained in candidate-accessible materials before the offer stage. For public companies, it includes whether per-level grant data goes beyond what SEC filings disclose.
Internal Pay Equity Data Published (20 points) measures whether companies voluntarily publish gender pay gap analysis, racial pay equity data, or equivalent, with methodology. UK-mandated gender pay gap filings and US EEO-1 submissions count as partial credit; proactive publication with intersectional methodology earns full marks.
Compensation Philosophy Public (15 points) measures whether the company's approach to compensation — market percentile target, cash-to-equity ratio, how refreshes work — is documented in a candidate-accessible format on its careers page or in public documents.
Market-Rate Commitment / External Benchmarking (10 points) measures whether the company publicly commits to a specific market percentile or publishes data that enables independent salary benchmarking.
Who Leads — and Why
The AA tier contains just two companies in this first edition.
Hugging Face (85, AA) leads the index by 3 points. The company publishes salary ranges on the overwhelming majority of its global job postings — including roles in France, Germany, Singapore, and the United Kingdom — without being legally required to do so. The transparency extends further. The company's engineering blog explains how startup equity works: vesting cliffs, dilution mechanics, and how grant values are set relative to funding round pricing. Most pre-IPO companies treat this information as NDA material until the offer stage. Hugging Face treats it as candidate-facing content. Its 9/10 Market-Rate Commitment score reflects specific language in careers materials about competitive benchmarking against European and North American tech markets. The primary gap to a hypothetical AAA score is the absence of a formal, published intersectional pay equity audit with methodology.
Stripe (82, AA) demonstrates that sustained institutional commitment to compensation visibility predates most of the US state laws now mandating salary range disclosure. Stripe posts bands on virtually all global job listings and maintains a careers page section — "How we think about compensation" — that goes beyond range disclosure to explain equity vesting schedules, annual review cadence, and the philosophy behind its bonus structure. Stripe's equity transparency score (18/25) reflects public materials that explain RSU mechanics and the relationship between private market valuations and equity grant refresh cycles in terms a candidate can act on. Its pay equity dimension (14/20) reflects an annual diversity report that includes gender pay gap data in independently auditable form.
The A tier — GitHub (75) and Microsoft (73) — reflects two different routes to similar scores. GitHub's strength is posting transparency: it extends range disclosure to most global postings beyond US state minimums. Microsoft's edge is pay equity infrastructure: its annual pay equity report, filed with institutional ESG investors and published on its DEI portal, covers gender and racial pay equity with significance testing. It is the most rigorous voluntary pay equity disclosure in the index. Neither earns above 80 because equity grant ranges and compensation philosophy documentation remain generic rather than candidate-specific.
The Middle Ground
Four companies occupy the BBB tier: Shopify (68), OpenAI (65), Anthropic (62), and Google DeepMind (60).
Shopify (68) performs creditably, benefiting from Canada's provincial pay transparency legislation which has influenced its posting practices beyond what US state law requires. Its annual Impact Report includes pay equity language with more specificity than most BBB-tier employers provide.
OpenAI (65) presents the most notable contrast in the index. The company that generated more public discourse about AI compensation than any other in 2023–24 does not systematically disclose salary ranges on global postings. Ranges appear in California, Colorado, New York, and Washington job postings — all driven by state law — but not on UK or European roles. The post-restructuring equity conversion from profit participation units to restricted stock units in 2025 introduced a consequential change to total compensation structure that was not documented in candidate-accessible public materials. Levels.fyi submissions and recruiter data remain the primary source of OpenAI compensation intelligence for external candidates.
Anthropic (62) operates in a similar posture. Salary ranges appear in state-compliant US postings but UK, European, and Asia-Pacific postings regularly omit them. The pre-IPO RSU structure is described generically in careers materials, without the grant-range specificity or vesting timeline detail that higher-ranked employers provide. The company scores 14/20 on pay equity, reflecting internal equity philosophy language in careers materials that does not yet constitute a published methodology.
Google DeepMind (60) illustrates the institutional paradox of large-company AI divisions: Alphabet's annual DEI report is substantive and covers DeepMind employees, but DeepMind-specific disclosures are thin. Global postings in India, Singapore, and Canada consistently carry no salary ranges in the July 2026 audit window.
The Disclosure Laggards
Twelve of the 20 companies in this index score B or BB. They include some of the sector's most prominent and highest-funded employers.
The BB tier — Databricks (58), Cohere (56), Meta AI (55), NVIDIA (53), Amazon AWS AI (52), and Scale AI (51) — reflects the modal behavior of large to mid-size AI employers: compliance with legal minimums, some diversity reporting, and little proactive voluntary disclosure. Meta AI (55) is the most instructive case. The company meets UK mandatory gender pay gap reporting requirements and posts ranges in required US states, but has taken no voluntary steps beyond legal thresholds. For a company with public commitments to pay equity, the disclosure posture in job postings is below what those commitments suggest.
The B tier anchors at 49 and descends to 30. Weights & Biases (49) comes closest to the BB threshold, posting ranges on most US job listings — a positive signal at its funding stage — but formal pay equity reporting is absent. xAI (40) sits near the bottom. Job postings on X carry minimal salary information, and Elon Musk's repeated public claims that xAI pays "the best in the industry" function as a general market commitment claim but equity terms for a private company at its reported valuation are entirely opaque to external candidates. Modal (30) and Replicate (35) anchor the index floor, consistent with early-stage infrastructure startup norms where most hiring occurs through networks rather than public postings — but the effect for candidates is identical to deliberate opacity.
What Comes Next
Two forces will reshape this index in H2 2026 and into 2027.
The EU Pay Transparency Directive — requiring member state implementation by June 2026, with enforcement penalties from 2027 — is already producing behavioral change at AI companies with European operations. Google DeepMind's London and Zurich postings, Mistral's Paris roles, and any US-headquartered AI employer posting in Germany, France, or the Netherlands will face enforceable disclosure requirements that include salary band publication and proactive pay equity reporting. Companies that built compliance infrastructure for US state mandates will find EU compliance incremental. Those that have not will face a material compliance buildout on a short timeline.
The second force is candidate behavior. In ENTRA's Q2 2026 Talent Survey (n=1,840 AI practitioners), 68% said they were less likely to advance in a hiring process where no salary range was visible in the job posting. That is a candidate behavior no compensation philosophy can override for a material share of the available talent pool. The AAA tier — no occupants in this first edition — requires full global posting disclosure, detailed equity transparency, an intersectional pay equity audit with methodology, a specific market percentile commitment, and a compensation philosophy document accessible before first-round interviews. Two companies in this index are within striking distance. The January 2027 refresh will test whether the first EU enforcement quarter accelerates the field.
How we ranked
The Top 20 AI Companies by Pay Transparency Score — August 2026 is scored across 5 dimensions:
- Salary Band Disclosure (30%) — Whether companies publish salary ranges in job postings across all geographies, not only legally mandated jurisdictions. (Source: ENTRA Pay Transparency Audit Q2 2026, n=87 companies screened; editorial review of live postings July 2026)
- Equity Compensation Transparency (25%) — Public disclosure of equity grant ranges, vesting schedules, dilution policy, and equity mechanics in candidate-accessible materials. (Source: Company careers pages, public blog posts, SEC filings, Levels.fyi Q2 2026)
- Internal Pay Equity Data Published (20%) — Publication of gender pay gap analysis, racial pay equity data, or equivalent with methodology — including UK-mandated GPG filings, US EEO-1 submissions, and voluntary DEI pay equity reports. (Source: ENTRA audit of company DEI portals and annual reports, Q1–Q2 2026)
- Compensation Philosophy Public (15%) — Presence of documented compensation philosophy in candidate-accessible form on careers page, company blog, or public document. (Source: ENTRA career-page audit Q2 2026)
- Market-Rate Commitment / External Benchmarking (10%) — Public commitment to a specific market percentile or publication of data enabling external salary benchmarking. (Source: Careers pages, public blog posts, press statements)
Data window: January 1 – June 30, 2026 (pay equity reports, DEI filings, annual reports); May 1 – July 20, 2026 (job posting audit, careers page review) Sample size: ENTRA Pay Transparency Audit Q2 2026 (n=87 AI companies screened; 20 scored for this index); approximately 120 live postings audited across 20 companies, July 2026; Levels.fyi Q2 2026 company-level disclosure data
Limitations:
- Private companies (Anthropic, Databricks, Cohere, Together AI, Replicate, Modal, Scale AI, xAI, Mistral, Weights & Biases) cannot be scored on SEC filing disclosures; equity transparency scores for private companies reflect only candidate-facing materials, which systematically understate disclosure available post-offer
- Salary band disclosure scores reflect job postings active during July 2026; companies with recent policy changes expanding range disclosure may score higher in the next refresh
Inquiries about methodology: methodology@entracareers.com