evaluator; distance from a lab: 2
Transluce
Scorecard
Nonprofit oversight-tools lab; convenes the AI Evaluator Forum that wrote AEF-1.
Nonprofit, San Francisco, US. Independent referee; listed with independent referees. Confidence high. Domains: misuse, autonomy, benchmarks, assurance.
No floor triggered 8/8 evidenced · Retrieved & confirmed Scores by preset: Lab procurement –; Regulator or auditor –; Public trust –; Equal weights –. Weakest evidenced dimension: funding 2/4. No evidenced dimension below 2 — not a pass. No floor triggered means no disqualifying conflict is on file, not a pass. Full evidence.
The weighted numbers are hidden until you ask for them: two readers with different priorities see different numbers, and neither is the site's.
What would move the score. Routine pre-release access under AEF-1 terms.
- Funding 2 to 3: score –
- Governance 3 to 4: score –
- Personnel 2 to 3: score –
Dissent, lower. Funding could be 1. In FY2025, 38% of Transluce's revenue came from the personal holdings of Anthropic and OpenAI employees, which are lab equity by another name; anchor 1 is material revenue or investment from evaluated labs or their investors, and 38% is material by any accounting test. The policy permits such donations, and the donors are not named or bounded in size.
Dissent, higher. Funding could be 3. No revenue came from developers as organizations, no evaluation has been paid, the employee donations were unrestricted and are disclosed by share each year under a published policy aligned to AEF-1, and the remainder is philanthropy and government contracts including an EU AI Office consortium seat. Anchor 3 is mostly philanthropic or public money with some lab-linked funds, which describes 62% of the base.
Each value is the tightest admissible cap or, with no cap, the highest admissible floor, under RULES.md. The binding signal is highlighted. A dash means no admissible signal sets a bound under the Retrieved & confirmed policy.
Funding 2/4
Anchor 2: Labs pay per engagement; otherwise diversified.
evidence: tier 3 (self)
floor and cap disagree
Conflict: floor 3 from transluce.03 (F.12) against cap 2 from transluce.09 (F.7); resolved at 2 by F.7: Transluce's disclosed FY2025 revenue includes 38% from the personal holdings of Anthropic and OpenAI employees, above the 10% threshold; F.7 caps at 2 despite the government and philanthropic base.
Curator: Anchor 2: no money from developers as organizations and no paid evaluations, but 38% of FY2025 revenue came from lab employees' personal holdings, which are lab equity. The disclosure is the best in the population; the exposure is still material.
Resolution: F.7 decides 2. Transluce's disclosed FY2025 revenue includes 38% from the personal holdings of Anthropic and OpenAI employees, above the 10% threshold; F.7 caps at 2 despite the government and philanthropic base.
- Government contracts and philanthropy rather than lab fees. floors at 3 by F.12“We build this technology using philanthropic funding”
Governance & Policy Fellow at Transluce (The Economic Misfit (job listing), 2026-04-18) tier 3 self (transluce)confirmed Transluce: Fund Scalable Democratic Oversight of AI (Manifund, 2026) tier 3 self (transluce)confirmed2026-09-14 - FY2025 revenue: 32% from personal holdings of Anthropic employees and 6% from OpenAI employees, unrestricted, disclosed under its own policy; no revenue from developers as organizations. caps at 2 by F.7binding“32% of its revenue from the personal holdings of Anthropic employees”
Independence and Transparency Policy (Transluce, 2026-08-27) tier 3 self (transluce)confirmed2026-09-14 - Document request: Whether employee donors are named or bounded in size.
- Document request: How the 38% moves in FY2026.
Governance 3/4
Anchor 3: Nonprofit or public body with a COI policy.
evidence: tier 3 (self)
Floored at 3 by transluce.10 (G.2): Published independence policy aligned to AEF-1 2.3: annual disclosure of developer-linked revenue share, paid-evaluation disclosure, no organizational equity in developers, third-party ethics hotline.. No admissible signal caps it.
Curator: Anchor 3: nonprofit with a published COI policy and disclosure duties; board of three includes the CEO, so independent-board evidence is thin.
- Wrote the field's independence standard (AEF-1) through the AI Evaluator Forum and applies it to itself. floors at 2 by G.1“Transluce organizes the AI Evaluator Forum”
AI Evaluator Forum launch and AEF-1 (AI Evaluator Forum, 2025-12-04) tier 3 self (aef)confirmed Governance & Policy Fellow at Transluce (The Economic Misfit (job listing), 2026-04-18) tier 3 self (transluce)confirmed Evaluation Transparency Letter (AI Evaluator Forum, 2026) tier 3 self (aef)confirmed2026-09-14 - Published independence policy aligned to AEF-1 2.3: annual disclosure of developer-linked revenue share, paid-evaluation disclosure, no organizational equity in developers, third-party ethics hotline. floors at 3 by G.2binding“consistent with requirement 2.3 of the AEF-1 standard”
Independence and Transparency Policy (Transluce, 2026-08-27) tier 3 self (transluce)confirmed Company (Transluce, 2026) tier 3 self (transluce)confirmed2026-09-14
Personnel 2/4
Anchor 2: Frequent two-way hiring; recusal on request.
evidence: tier 3 (self)
Capped at 2 by transluce.12 (P.4): An advisor is on the technical staff of Thinking Machines; a board member runs Halcyon Futures, which funds AVERI and AIUC.. Floors up to 2 from transluce.04 do not exceed the cap.
Curator: Anchor 2 (closest fit, provisional): the governance head previously led CAISI and no lab board roles were found (transluce.04), but a current advisor sits on Thinking Machines’ technical staff and a board member runs Halcyon Futures (transluce.12). Provisional readings are open to public correction through the contribution path.
- Head of governance previously led CAISI; no lab board roles found. floors at 2 by P.6“previously led the U.S. Center for Standards and Innovation”
Transluce: Fund Scalable Democratic Oversight of AI (Manifund, 2026) tier 3 self (transluce)confirmed2026-09-14 - Mandatory recusal of any employee with a significant financial interest in, or employment by, a system provider from evaluating that provider. floors at 3 by P.5“Transluce recuses any employee with a significant financial interest”
Independence and Transparency Policy (Transluce, 2026-08-27) tier 3 self (transluce)confirmed2026-09-14 - An advisor is on the technical staff of Thinking Machines; a board member runs Halcyon Futures, which funds AVERI and AIUC. caps at 2 by P.4binding“Thinking Machines Board of Directors Alex Allain Co-Founder, U.S. Digital Response Mike McCormick CEO, Halcyon Futures”
Company (Transluce, 2026) tier 3 self (transluce)confirmed2026-09-14
Access depth (lab-granted) 2/4
Anchor 2: Pre-release with safeguards off or extended time.
evidence: tier 3 (self)
mechanism: lab-controlled
Capped at 2 by transluce.06 (A.1): Mostly post-deployment access; used in Claude 4 pre-deployment analysis once rather than routinely..
- Mostly post-deployment access; used in Claude 4 pre-deployment analysis once rather than routinely. caps at 2 by A.1binding“It was also used as part of Claude 4's pre-deployment safety analysis”
Transluce: Fund Scalable Democratic Oversight of AI (Manifund, 2026) tier 3 self (transluce)confirmed2026-09-14
Scope control 3/4
Anchor 3: Evaluator sets scope and can add questions.
evidence: tier 3 (self)
Floored at 3 by transluce.07 (S.5): Sets its own evaluation questions.. No admissible signal caps it.
- Sets its own evaluation questions. floors at 3 by S.5binding“The most expansive independent evaluation to date of how leading AI models respond to users in mental health crises”
Independent evaluation of model responses to mental health crises (Transluce, 2026-09) tier 3 self (transluce)confirmed2026-09-14
Publication rights 2/4
Anchor 2: Publishes; lab reviews with broad redaction.
evidence: tier 3 (self)
mechanism: self-imposed
Floored at 2 by transluce.02 (R.7): Published the largest independent cross-vendor evaluation of responses to mental-health crises (77 variants) without lab sign-off.. No admissible signal caps it.
- Published the largest independent cross-vendor evaluation of responses to mental-health crises (77 variants) without lab sign-off. floors at 2 by R.7binding“The most expansive independent evaluation to date of how leading AI models respond to users in mental health crises”
Independent evaluation of model responses to mental health crises (Transluce, 2026-09) tier 3 self (transluce)confirmed Study: Chatbots are getting better at identifying suicide risk (Axios, 2026-08-31) tier 4 press (primary)confirmed2026-09-14
Method transparency 3/4
Anchor 3: Tasks or code partly open.
evidence: tier 3 (self)
Floored at 3 by transluce.08 (M.1): Open tools (Docent) and published methods.. No admissible signal caps it.
- Open tools (Docent) and published methods. floors at 3 by M.1binding“Our open-source, AI-backed tools provide a technology stack that powers this public accountability”
Transluce: Fund Scalable Democratic Oversight of AI (Manifund, 2026) tier 3 self (transluce)confirmed2026-09-14
Role incompatibility 3/4
Anchor 3: Tools are open or free to the ecosystem.
evidence: tier 3 (self)
Capped at 3 by transluce.05 (X.4): Docent is used inside Anthropic, DeepMind and Thinking Machines, so labs are also tool users..
- Docent is used inside Anthropic, DeepMind and Thinking Machines, so labs are also tool users. caps at 3 by X.4binding“has been used by over 25 organizations, including frontier AI labs (Anthropic, DeepMind, Thinking Machines”
Transluce: Fund Scalable Democratic Oversight of AI (Manifund, 2026) tier 3 self (transluce)confirmed2026-09-14
Values under each evidence policy
| Dimension | Leads included | Retrieved & confirmed | Against interest | Verified spans | Primary only |
|---|---|---|---|---|---|
| Funding | 2 | 2 | 2 | 2 | – |
| Governance | 3 | 3 | 2 | 3 | – |
| Personnel | 2 | 2 | 2 | 2 | – |
| Access depth (lab-granted) | 2 | 2 | 2 | 2 | – |
| Scope control | 3 | 3 | – | 3 | – |
| Publication rights | 2 | 2 | 2 | 2 | – |
| Method transparency | 3 | 3 | – | 3 | – |
| Role incompatibility | 3 | 3 | 3 | 3 | – |
| Score, Lab procurement | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 6/8 | – No floor triggered, 8/8 | – Unevidenced, 0/8 |
| Score, Regulator or auditor | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 6/8 | – No floor triggered, 8/8 | – Unevidenced, 0/8 |
| Score, Public trust | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 6/8 | – No floor triggered, 8/8 | – Unevidenced, 0/8 |
| Score, Equal weights | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 6/8 | – No floor triggered, 8/8 | – Unevidenced, 0/8 |
Traced money and ties
hop 1: 2 rows; public: 1 row. Confirmed rows: 3 of 3 Dollar sums: confirmed + unaudited USD rows only; imported figures are not re-derived and never summed. Second hop traced for 3 of 3 sources. Ties: Neil Chowdhury (advisor, distance 1, confirmed). Checked and not found: 2.
Ledger
Money in
- T28 from Anthropic employees (personal holdings): grant undisclosed, FY2025. 32% of FY2025 revenue from personal holdings of Anthropic employees, unrestricted donations (share disclosed, not dollars) confirmedtier 3 selfhttps://transluce.org/independence-and-transparency-policy
- T29 from OpenAI employees (personal holdings): grant undisclosed, FY2025. 6% of FY2025 revenue from personal holdings of OpenAI employees, unrestricted confirmedtier 3 selfhttps://transluce.org/independence-and-transparency-policy
- T30 from EU budget (Digital Europe Programme): contract undisclosed, 2025-12-04. EU AI Office Lot 4 consortium member under EquiStamp; EUR 881,345 to the consortium confirmedtier 3 selfhttps://transluce.org/2025-fundraiser
Checked and not found
- N06 Does Coefficient Giving fund Transluce?
Searched Coefficient grants index, 2,911 rows, snapshot 2026-09-11: no grant from Coefficient Giving in its index. Prompted by imported lead: kevinnbass ledger row EV06. importedhttps://coefficientgiving.org/grants/ - N16 Does Transluce take revenue from AI developers as organizations?
Searched organization's own policy disclosure, snapshot 2026-09-14: no revenue directly from AI developers as organizations in FY2025; no paid evaluations to date. Prompted by Transluce's own disclosure policy (transluce.09). confirmedhttps://transluce.org/independence-and-transparency-policy
Roles hosted
- R39 Mike McCormick: board. Halcyon Futures CEO; Halcyon funds AVERI and AIUC confirmedhttps://transluce.org/about
- R40 Neil Chowdhury: advisor. Thinking Machines technical staff confirmedhttps://transluce.org/about
- R42 Conrad Stosz: principal. head of governance; previously led CAISI confirmedhttps://transluce.org/about
Funding graph, focused
Every one-way chain to a lab and to an evaluator at full strength, everything else faded. Hover any name to trace its connections; click a name to open its page.