evaluator; distance from a lab: 2
Apollo Research
Scorecard
Scheming and deception specialist; ran evaluations for every major lab; converted to a public benefit corporation in January 2026.
Public benefit corp, London, UK / San Francisco, US. Independent referee; listed with independent referees. Confidence high. Domains: scheming, autonomy, 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. Disclose revenue concentration by client and separate the product business from the evaluation practice.
- Funding 2 to 3: score –
- Governance 2 to 3: score –
- Personnel 2 to 3: score –
Dissent, lower. Funding could be 1. Apollo asks the parties it evaluates to pay fair market value, says it ran evaluations for all major labs in 2025, and publishes no revenue split; its 2024 self-report named only about $2M of philanthropy, and its 2026 seed round included Macroscopic Ventures, an Anthropic Series A and B investor. If lab fees are the principal revenue, F.5 caps at 1, and a material lab-investor stake would cap at 1 under F.4.
Dissent, higher. Funding could be 3. Apollo's norms bar outcome-contingent pay and side grants or investments from evaluated labs; the PBC post says it remains partly philanthropy-supported, and the index records about $4.4M from Coefficient and $1.1M from SFF. A published cap holding lab revenue at or below 10% of annual revenue, with market rates and publication rights, would lift the F.3 cap to 3, the anchor for mostly philanthropic money with some lab-linked pooled funds.
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 1 (filing/index)
Capped at 2 by apollo.02 (F.3): Requests fair-market-value compensation from the parties it evaluates; revenue depends on evaluated labs.; capped at 2 by apollo.15 (F.4): Macroscopic Ventures, an investor in Anthropic's Series A and B, participated in Apollo's January 2026 seed round; 50Y led and no stake sizes were disclosed.. Floors up to 2 from apollo.12 do not exceed the cap. Not counted under this policy: apollo.11 (no confirmed source (imported)).
- Requests fair-market-value compensation from the parties it evaluates; revenue depends on evaluated labs. caps at 2 by F.3binding“we generally request that the parties for whom we conduct work compensate us at a fair market value for the work”
Our Norms on Security, Science Communication and Conflicts of Interest (Apollo Research, 2025-11-26) tier 3 self (apollo)confirmed2026-09-14 - SFF recommendations of about $1.1M and Coefficient grants of about $4.4M per the index snapshot; both funders are one to two steps from an Anthropic investor. caps at 3 by F.6not counted under standard
evaluators.csv (24 evaluators: Coefficient and SFF totals, lab money, government contracts, leadership) (Kevin Bass (GitHub), 2026-09-14) tier 2 ledgerimported Coefficient Giving grants index (Coefficient Giving, 2026-09-11) tier 1 indeximported2026-09-11 - Self-reported philanthropic base: about $1.5M from Open Philanthropy and about $500K from SFF at the time of its Manifund listing; a later index total of $4.41M is recorded as an imported row and the two figures are not reconciled. floors at 2 by F.5“We have received ~$1.5M from OpenPhil and ~$500k from SFF”
Apollo Research on Manifund (Manifund, 2024) tier 3 self (apollo)confirmed evaluators.csv (24 evaluators: Coefficient and SFF totals, lab money, government contracts, leadership) (Kevin Bass (GitHub), 2026-09-14) tier 2 ledgerimported2026-09-15 - Macroscopic Ventures, an investor in Anthropic's Series A and B, participated in Apollo's January 2026 seed round; 50Y led and no stake sizes were disclosed. caps at 2 by F.4binding“further investments from Juniper Ventures , Macroscopic Ventures”
Apollo Research is becoming a PBC (investment section) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed Macroscopic Ventures investor profile (Premier Alts, 2026) tier 4 press (aggregator)confirmed2026-09-15 - Coefficient / Open Philanthropy recommended $1,535,480 (June 2023, startup costs) and $2,178,700 (May 2024, general support) to Apollo per archived grant pages, with a further $696,000 (May 2026) in the removed index; Coefficient's principal funder is an Anthropic investor. caps at 3 by F.6“Open Philanthropy recommended a grant of $2,178,700 to Apollo Research for general support.”
Apollo Research — Startup Funding (Wayback capture 2023-09-12 of the Open Philanthropy grant page) (Open Philanthropy via Internet Archive, 2023-06) tier 1 indexconfirmed Apollo Research — General Support (Wayback capture 2024-07-17 of the Open Philanthropy grant page) (Open Philanthropy via Internet Archive, 2024-05) tier 1 indexconfirmed Apollo Research on Manifund (Manifund, 2024) tier 3 self (apollo)confirmed Wayback CDX index of openphilanthropy.org/grants/a* (status 200, one row per URL) (Internet Archive, 2026-09-15) tier 1 indexconfirmed2026-09-15 - The January 2026 seed round's size and each investor's stake, including Macroscopic Ventures' (an Anthropic Series A and B investor), are undisclosed: no Form D in EDGAR, no share allotment at Companies House, and the PBC is a Delaware entity with no public cap table. caps at 3 by F.10“The round was oversubscribed.”
Apollo Research is becoming a PBC (investment section) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed EDGAR full-text search: "Apollo Research", form D, 2025-06-01 to 2026-09-15 (U.S. Securities and Exchange Commission, 2026-09-15) tier 1 indexconfirmed APOLLO RESEARCH AI LTD filing history (Companies House, 2026) tier 1 filingconfirmed APOLLO RESEARCH AI LTD persons with significant control (Companies House, 2026) tier 1 filingconfirmed APOLLO RESEARCH AI LTD (company 15289159), overview (Companies House, 2026) tier 1 filingconfirmed MR01 Registration of charge 152891590001 (PDF) (Companies House, 2025-12-10) tier 1 filingconfirmed2026-09-15 - Document request: Reconcile the self-reported $1.5M with the index total of $4.41M by date.
Governance 2/4
Anchor 2: For-profit or PBC with a published COI policy.
evidence: tier 1 (filing/index)
Capped at 2 by apollo.03 (G.2): Converted from fiscally sponsored nonprofit to a public benefit corporation in January 2026.; capped at 2 by apollo.10 (G.3): January 2026 seed round led by 50Y with Juniper, Macroscopic Ventures (an Anthropic Series A and B investor), Common Metal, SAIF and individuals; amount not stated. Mission board seats designated as a governance safeguard.; capped at 2 by apollo.14 (G.3): Macroscopic Ventures, a participant in the seed round, lists Anthropic Series A and Series B among its investments.. Floors up to 2 from apollo.01, apollo.13 do not exceed the cap.
Curator: Anchor 2: for-profit PBC with a published COI policy and designated mission seats; investors include a lab investor. Previously mis-scored at 3 as if nonprofit.
- Published four-rule COI policy: no outcome-contingent pay, no side grants or investments from evaluated labs, recusal for financial interests, no misrepresentation. floors at 2 by G.2“We do not accept work where compensation is contingent upon the outcome”
Our Norms on Security, Science Communication and Conflicts of Interest (Apollo Research, 2025-11-26) tier 3 self (apollo)confirmed2026-09-14 - Converted from fiscally sponsored nonprofit to a public benefit corporation in January 2026. caps at 2 by G.2binding“Apollo is spinning off from our fiscal sponsor into a Public Benefit Corporation (PBC)”
Apollo Research is becoming a PBC (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed2026-09-14 - January 2026 seed round led by 50Y with Juniper, Macroscopic Ventures (an Anthropic Series A and B investor), Common Metal, SAIF and individuals; amount not stated. Mission board seats designated as a governance safeguard. caps at 2 by G.3binding“raised a Seed round led by 50Y”
Apollo Research is becoming a PBC (investment section) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed2026-09-14 - Mission seats on the PBC board are held by directors independent of the company and its funders, with a mandate to prioritize mission. floors at 2 by G.2“who are independent to Apollo Research PBC and our funders”
Apollo Research is becoming a PBC (governance) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed2026-09-15 - Macroscopic Ventures, a participant in the seed round, lists Anthropic Series A and Series B among its investments. caps at 2 by G.3binding“further investments from Juniper Ventures , Macroscopic Ventures”
Macroscopic Ventures investor profile (Premier Alts, 2026) tier 4 press (aggregator)confirmed Apollo Research is becoming a PBC (investment section) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed2026 - The PBC board has three seats, of which the 3rd (and later 6th) are designated mission seats; only the first mission director, Daniel Kokotajlo, is named, and the remaining directors are not published on Apollo's pages or in Companies House (which shows only the UK subsidiary's directors, Hobbhahn and Akin). caps at 3 by G.8“Our first mission director is Daniel Kokotajlo.”
Apollo Research is becoming a PBC (investment section) (Apollo Research, 2026-01-20) tier 3 self (apollo)confirmed APOLLO RESEARCH AI LTD officers (Companies House, 2026) tier 1 filingconfirmed Team (Apollo Research, 2026) tier 3 self (apollo)confirmed2026-09-15 - Document request: Round size and Macroscopic's share.
- Document request: Who holds the mission seats.
Personnel 2/4
Anchor 2: Frequent two-way hiring; recusal on request.
evidence: tier 3 (self)
Floored at 2 by apollo.05 (P.5): Recusal rule for any individual with a financial interest in an evaluated organization.. No admissible signal caps it.
- Recusal rule for any individual with a financial interest in an evaluated organization. floors at 2 by P.5binding“We recuse any individual who is considered to have a financial interest in an organization we are evaluating”
Our Norms on Security, Science Communication and Conflicts of Interest (Apollo Research, 2025-11-26) tier 3 self (apollo)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
Floored at 2 by apollo.06 (A.1): Ran pre-deployment evaluations for all major labs; contracted by UK AISI for deception evaluations.. No admissible signal caps it.
- Ran pre-deployment evaluations for all major labs; contracted by UK AISI for deception evaluations. floors at 2 by A.1binding“partnered with OpenAI to test their o1 model before public deployment”
About (Apollo Research, 2026) tier 3 self (apollo)confirmed The First Year of Apollo Research (Apollo Research, 2024) tier 3 self (apollo)confirmed2026-09-14
Scope control 3/4
Anchor 3: Evaluator sets scope and can add questions.
evidence: tier 3 (self)
Capped at 3 by apollo.07 (S.6): Co-authored anti-scheming research with OpenAI while acting as its external scheming evaluator..
- Co-authored anti-scheming research with OpenAI while acting as its external scheming evaluator. caps at 3 by S.6binding“partnered with OpenAI to study anti-scheming interventions on frontier models”
About (Apollo Research, 2026) tier 3 self (apollo)confirmed2026-09-14
Publication rights 2/4
Anchor 2: Publishes; lab reviews with broad redaction.
evidence: tier 3 (self)
mechanism: unknown
Floored at 2 by apollo.04 (R.7): Record of adverse findings published, including in-context scheming and evaluation awareness.. No admissible signal caps it.
- Record of adverse findings published, including in-context scheming and evaluation awareness. floors at 2 by R.7binding“published the first evidence that frontier models can scheme in context”
About (Apollo Research, 2026) tier 3 self (apollo)confirmed2026-09-14
Method transparency 2/4
Anchor 2: Methods described in prose.
evidence: tier 3 (self)
Floored at 2 by apollo.09 (M.2): Publishes evaluation methodology and papers; differential publishing policy documented.. No admissible signal caps it.
- Publishes evaluation methodology and papers; differential publishing policy documented. floors at 2 by M.2binding“We engage in differential publishing”
Our Norms on Security, Science Communication and Conflicts of Interest (Apollo Research, 2025-11-26) tier 3 self (apollo)confirmed2026-09-14
Role incompatibility 2/4
Anchor 2: Consults for labs.
evidence: tier 3 (self)
Capped at 2 by apollo.08 (X.1): Sells Watcher, a monitoring product, into the ecosystem it evaluates..
- Sells Watcher, a monitoring product, into the ecosystem it evaluates. caps at 2 by X.1binding“building an AI security tool to monitor frontier AI agents”
About (Apollo Research, 2026) tier 3 self (apollo)confirmed2026-09-14
Values under each evidence policy
| Dimension | Leads included | Retrieved & confirmed | Against interest | Verified spans | Primary only |
|---|---|---|---|---|---|
| Funding | 2 | 2 | 2 | 2 | 3 |
| Governance | 2 | 2 | 2 | 2 | 3 |
| Personnel | 2 | 2 | – | 2 | – |
| Access depth (lab-granted) | 2 | 2 | – | 2 | – |
| Scope control | 3 | 3 | 3 | 3 | – |
| Publication rights | 2 | 2 | – | 2 | – |
| Method transparency | 2 | 2 | – | 2 | – |
| Role incompatibility | 2 | 2 | 2 | 2 | – |
| Score, Lab procurement | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 4/8 | – No floor triggered, 8/8 | – No floor triggered, 2/8 |
| Score, Regulator or auditor | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 4/8 | – No floor triggered, 8/8 | – No floor triggered, 2/8 |
| Score, Public trust | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 4/8 | – No floor triggered, 8/8 | – No floor triggered, 2/8 |
| Score, Equal weights | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 4/8 | – No floor triggered, 8/8 | – No floor triggered, 2/8 |
Traced money and ties
hop 1: 1 row; hop 2: 3 rows (recommendation $3.7M); unattributed: 1 row. Confirmed rows: 4 of 5 (2 excluded after a re-fetch disagreed or could not be verified: T08, T33) Dollar sums: confirmed + unaudited USD rows only; imported figures are not re-derived and never summed. Second hop traced for 2 of 3 sources: untraced Fifty Years (50Y). Ties: Macroscopic Ventures (formerly Center for Emerging Risk Research) (investor, distance 1, confirmed). Checked and not found: 1.
Ledger
Money in
- T08 from Coefficient Giving (formerly Open Philanthropy): grant $4.41M; quarantined — excluded from sums, 2021-2025. cumulative unverifiabletier 1 indexhttps://coefficientgiving.org/
- T31 from Fifty Years (50Y): investment undisclosed, 2026-01-20. seed round led by 50Y with Juniper, Macroscopic, Common Metal, SAIF, Ocean, Progress Fund, individuals; amount not stated confirmedtier 3 selfhttps://www.apolloresearch.ai/blog/apollo-research-is-becoming-a-pbc
- T32 from Macroscopic Ventures (formerly Center for Emerging Risk Research): investment undisclosed, 2026-01-20. participant in the seed round confirmedtier 3 selfhttps://www.apolloresearch.ai/blog/apollo-research-is-becoming-a-pbc
- T33 from Survival and Flourishing Fund: recommendation $1.13M; quarantined — excluded from sums, 2023-2025. SFF recommendations unverifiabletier 3 selfhttps://survivalandflourishing.fund/recommendations
- T85 from Coefficient Giving (formerly Open Philanthropy): recommendation $1.54M, 2023-06. startup costs (Open Philanthropy grant page, archived) confirmedtier 1 indexhttps://web.archive.org/web/20230912095624/https://www.openphilanthropy.org/grants/apollo-research-startup-funding/
- T86 from Coefficient Giving (formerly Open Philanthropy): recommendation $2.18M, 2024-05. general support (Open Philanthropy grant page, archived) confirmedtier 1 indexhttps://web.archive.org/web/20240717140504/https://www.openphilanthropy.org/grants/apollo-research-general-support/
- T87 from Coefficient Giving (formerly Open Philanthropy): recommendation $0.70M; imported figure, not re-derived — never summed, 2026-05. staff support, per the third-party Algolia extract of Coefficient's index; no public page was ever captured importedtier 2 ledgerhttps://raw.githubusercontent.com/kevinnbass/metr-money-figure/master/research/evaluators.csv
Checked and not found
- N25 no Form D or other filing naming Apollo Research in EDGAR full-text search (2025-01-01 to 2026-09-15); no return of allotment (SH01) at Companies House for Apollo Research AI Ltd after its GBP 0.01 incorporation capital; PSC register states no registrable person
Searched SEC EDGAR full-text search (forms D and unrestricted) and company-name search; Companies House filing history and PSC register for company 15289159 through 2026-04-10, snapshot 2026-09-15: no Form D or other filing naming Apollo Research in EDGAR full-text search (2025-01-01 to 2026-09-15); no return of allotment (SH01) at Companies House for Apollo Research AI Ltd after its GBP 0.01 incorporation capital; PSC register states no registrable person. confirmedhttps://efts.sec.gov/LATEST/search-index?q=%22Apollo%20Research%22&forms=D&dateRange=custom&startdt=2025-06-01&enddt=2026-09-15
Roles hosted
- R37 Macroscopic Ventures (formerly Center for Emerging Risk Research): investor, 2026-01. supersedes R29 confirmedhttps://www.apolloresearch.ai/blog/apollo-research-is-becoming-a-pbc
- R38 Fifty Years (50Y): investor, 2026-01. lead confirmedhttps://www.apolloresearch.ai/blog/apollo-research-is-becoming-a-pbc
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.