evaluator; distance from a lab: unattributed
EU AI Office
Scorecard
Regulator behind the GPAI Code of Practice; external-evaluator obligations enforceable since August 2026.
Government, Brussels, BE. Government institute; listed with government institutes. Confidence med. Domains: bio, cyber, misuse, assurance.
No floor triggered 8/8 evidenced · Retrieved & confirmed Scores by preset: Lab procurement –; Regulator or auditor –; Public trust –; Equal weights –. Weakest evidenced dimension: publication rights 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. An accreditation pathway for external evaluators and a public reporting schema.
- Funding 3 to 4: score –
- Governance 3 to 4: score –
- Personnel 3 to 4: score –
Dissent, lower. Scope (S) at 2 could be 1. The public record shows the Office has run no evaluation of any frontier model; the only scope it has exercised is a procurement of technical assistance whose deliverables are shaped by the Code that signatory labs themselves negotiated. Until the Office uses Article 92, the labs effectively define what is tested through their own Safety and Security Model Reports, which is the anchor-1 picture.
Dissent, higher. Scope (S) at 2 could be 3. Article 92 gives the Office the legal power to evaluate any general-purpose model and to compel access, including source code, with fines for refusal, and enforcement has been live since August 2026. The Office set the six risk lots of the tender itself. A body that writes the questions and can compel answers is at least anchor 3; section 9 holds it down only for lack of a published run.
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 3/4
Anchor 3: Mostly philanthropic or public money; some lab-linked pooled funds.
evidence: tier 1 (filing/index)
Capped at 3 by euaio.10 (F.12): Public budget under the Digital Europe Programme; EUR 7.37M awarded across six lots to contracted evaluators. No confirmed negative on file.. Floors up to 3 from euaio.06, euaio.14 do not exceed the cap.
Curator: Anchor 3 pending a confirmed bounded negative; funding is public by construction but the gate asks for the row.
- Public funding. floors at 3 by F.9“established within the Commission as part of the administrative structure of the Directorate-General”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed Commission Decision Establishing the European AI Office (European Commission (DG CONNECT), 2024-01-24) tier 3 self (euaio)confirmed2026-09-14 - Public budget under the Digital Europe Programme; EUR 7.37M awarded across six lots to contracted evaluators. No confirmed negative on file. caps at 3 by F.12binding“The €9,080,000 tender is divided into six lots”
TED 864574-2025 Technical Assistance for AI Safety (Tenders Electronic Daily, 2025-12) tier 1 filingconfirmed The EU AI Act Newsletter #77: AI Office Tender (EU AI Act Newsletter (Substack), 2025-05) tier 4 press (aggregator)confirmed2026-09-14 - Commission Decision C(2024) 390 Article 8 funds the AI Office from DG CONNECT staff, Digital Europe Programme administrative appropriations, and DEP Specific Objective 2 'Artificial Intelligence' for operational expenditure, naming no non-Union source. floors at 3 by F.9“Operational expenditure of the Office shall be covered by the financial resources allocated to Specific Objective 2”
Commission Decision of 24 January 2024 establishing the European Artificial Intelligence Office, C/2024/1459 (Official Journal PDF) (Publications Office of the European Union, EUR-Lex, 2024-02-14) tier 1 filingconfirmed Commission Decision Establishing the European AI Office (European Commission (DG CONNECT), 2024-01-24) tier 3 self (euaio)confirmed2026-09-15 - Document request: Add a confirmed bounded negative from a primary source to restore anchor 4.
Governance 3/4
Anchor 3: Nonprofit or public body with a COI policy.
evidence: tier 1 (filing/index)
Floored at 3 by euaio.12 (G.2): AI Office staff are Commission officials bound by the EU Staff Regulations, which require screening of actual or potential conflicts of interest (Article 11a) and restrict post-employment activity for two years (Article 16).. No admissible signal caps it.
Curator: Anchor 3: public body with a statutory basis (euaio.07). No tier-1 source for an independent board or external review, so 4 is unearned under the tier rule (C10). Whether a formal COI policy is published is an open question.
- Regulator with statutory basis. floors at 2 by G.1“established within the European Commission as the foundation for a single AI governance system”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed European AI Office (European Commission (DG CONNECT), 2026) tier 3 self (euaio)confirmed2026-09-14 - AI Office staff are Commission officials bound by the EU Staff Regulations, which require screening of actual or potential conflicts of interest (Article 11a) and restrict post-employment activity for two years (Article 16). floors at 3 by G.2binding“a concrete procedure for properly screening new staff for actual or potential conflicts of interest”
New EU Staff Regulations adopted: Small steps on revolving door, giant leaps still needed (Corporate Europe Observatory, 2013-07) tier 4 press (aggregator)confirmed Staff Regulations of Officials of the European Union (consolidated text) (EUR-Lex, 2024-01-01) tier 1 filingunaudited2026-09-15
Personnel 3/4
Anchor 3: Recusal policy and disclosure of lab ties.
evidence: tier 1 (filing/index)
Floored at 3 by euaio.08 (P.5): Civil-service conflict rules.. No admissible signal caps it.
- Civil-service conflict rules. floors at 3 by P.5binding“Article 16 of the old staff regulations included a two-year notification period after staff leave their EU job”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed New EU Staff Regulations adopted: Small steps on revolving door, giant leaps still needed (Corporate Europe Observatory, 2013-07) tier 4 press (aggregator)confirmed Staff Regulations of Officials of the European Union (consolidated text) (EUR-Lex, 2024-01-01) tier 1 filingunaudited2026-09-14
Access depth (lab-granted) 3/4
Anchor 3: Helpful-only or weights-level access, chain of thought, logs, on-site.
evidence: tier 1 (filing/index)
mechanism: statutory
Floored at 3 by euaio.01 (A.2): Legal power to compel; the only body whose access does not depend on goodwill.. No admissible signal caps it.
- Legal power to compel; the only body whose access does not depend on goodwill. floors at 3 by A.2binding“may request access to the general-purpose AI model concerned through APIs or further appropriate technical means”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed Article 92: Power to conduct evaluations (AI Act Service Desk) (European Commission (AI Act Service Desk), 2024-07-12) tier 1 filingconfirmed2026-09-14
Scope control 4/4
Anchor 4: Evaluator sets scope, can investigate incidents, can refuse sign-off.
evidence: tier 1 (filing/index)
Floored at 4 by euaio.13 (S.4): The AI Office holds the legal power under the AI Act to evaluate general-purpose models with systemic risk and to compel access and information; scope for those evaluations is set by the regulator, not the developer.. No admissible signal caps it.
- Contracts its own evaluators (FAR.AI-led consortium) rather than relying on lab-chosen ones. floors at 3 by S.3“FAR AI was selected by the European Commission's AI Office to lead CBRN risk research under tender EC-CNECT/2025/OP/0032”
FAR AI (Longterm Wiki, 2026-02-26) tier 4 press (aggregator)confirmed2026-09-14 - Code sets the access floor: helpful-only versions where security allows, at least 20 business days, external evaluators for each new frontier model. floors at 3 by S.4“A developer should engage independent external evaluators for each new frontier model”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed Expanding External Access to Frontier AI Models for Dangerous Capability Evaluations (arXiv 2601.11916) (arXiv, 2026-01-21) tier 4 press (primary)confirmed2026-09-14 - Contract notice TED 864574-2025, Technical Assistance for AI Safety, published 26 December 2025; contractors self-report include METR, FAR.AI, SecureBio, SaferAI and Epoch. floors at 3 by S.3“This tender has been split into six lots”
TED notice 864574-2025: Artificial Intelligence Act: Technical Assistance for AI Safety (Tenders Electronic Daily, 2025-12-26) tier 1 filingconfirmed TED 864574-2025 Technical Assistance for AI Safety (Tenders Electronic Daily, 2025-12) tier 1 filingconfirmed Forthcoming call for tenders: Artificial Intelligence Act - Technical Assistance for AI Safety (European Commission (DG CONNECT), 2025-04-28) tier 3 self (euaio)confirmed The EU AI Act Newsletter #77: AI Office Tender (EU AI Act Newsletter (Substack), 2025-05) tier 4 press (aggregator)confirmed2025-12-26 - The AI Office holds the legal power under the AI Act to evaluate general-purpose models with systemic risk and to compel access and information; scope for those evaluations is set by the regulator, not the developer. floors at 4 by S.4binding“may request access to the general-purpose AI model concerned through APIs or further appropriate technical means”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed Article 92: Power to conduct evaluations (AI Act Service Desk) (European Commission (AI Act Service Desk), 2024-07-12) tier 1 filingconfirmed2026-09-15
Publication rights 2/4
Anchor 2: Publishes; lab reviews with broad redaction.
evidence: tier 3 (self)
mechanism: statutory
Capped at 2 by euaio.04 (R.4): Safety and Security Model Reports are submitted to the Office, not the public..
- Safety and Security Model Reports are submitted to the Office, not the public. caps at 2 by R.4binding“Signatories commit to reporting to the AI Office information about their model”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed General-Purpose AI Code of Practice: Safety and Security chapter (mirror) (code-of-practice.ai (Alexander Zacherl mirror of the Commission text), 2025-07) tier 4 press (aggregator)confirmed2026-09-14
Method transparency 2/4
Anchor 2: Methods described in prose.
evidence: tier 4 (press)
Capped at 2 by euaio.05 (M.2): No accreditation pathway or public reporting schema yet; thin evaluator bench..
- No accreditation pathway or public reporting schema yet; thin evaluator bench. caps at 2 by M.2binding“the pool of qualified evaluators remains thin”
The EU's Real AI Leverage Is Making Compliance the Path of Least Resistance (TechPolicy.Press, 2026-02-26) tier 4 press (primary)confirmed2026-09-14
Role incompatibility 4/4
Anchor 4: No commercial products.
evidence: tier 3 (self)
Floored at 4 by euaio.09 (X.7): No commercial products.. No admissible signal caps it.
- No commercial products. floors at 4 by X.7binding“It also enforces the rules for GPAI models and supports the governance bodies in Member States”
Frontier AI safety regulations: A reference for lab staff (METR, 2026-01-29) tier 3 self (metr)confirmed European AI Office (European Commission (DG CONNECT), 2026) tier 3 self (euaio)confirmed The EU AI Act Newsletter #77: AI Office Tender (EU AI Act Newsletter (Substack), 2025-05) tier 4 press (aggregator)confirmed2026-09-14
Values under each evidence policy
| Dimension | Leads included | Retrieved & confirmed | Against interest | Verified spans | Primary only |
|---|---|---|---|---|---|
| Funding | 3 | 3 | 3 | 3 | 3 |
| Governance | 3 | 3 | 3 | 3 | – |
| Personnel | 3 | 3 | 3 | 3 | – |
| Access depth (lab-granted) | 3 | 3 | 3 | 3 | 3 |
| Scope control | 4 | 4 | 4 | 4 | 4 |
| Publication rights | 2 | 2 | 2 | 2 | – |
| Method transparency | 2 | 2 | 2 | 2 | – |
| Role incompatibility | 4 | 4 | 4 | 4 | – |
| Score, Lab procurement | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 3/8 |
| Score, Regulator or auditor | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 3/8 |
| Score, Public trust | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 3/8 |
| Score, Equal weights | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 8/8 | – No floor triggered, 3/8 |
Traced money and ties
public: 1 row. Confirmed rows: 1 of 1 Dollar sums: confirmed + unaudited USD rows only; imported figures are not re-derived and never summed. Second hop traced for 1 of 1 sources. Ties: none within two steps recorded. Checked and not found: 2.
Ledger
Money in
- T48 from EU budget (Digital Europe Programme): contract €7.37M (EUR, not converted), 2025-12. EU AI Office: total awarded across six lots of TED 864574-2025 to contracted evaluators (outflow from the Office, recorded on the Office row for visibility) confirmedtier 1 filinghttps://ted.europa.eu/en/notice/-/detail/864574-2025
Checked and not found
- N10 Does Coefficient Giving fund the EU AI Office?
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 EV13. importedhttps://coefficientgiving.org/grants/ - N27 Commission Decision C(2024) 390 Article 8 names only Union budget sources for the AI Office: DG CONNECT staff, Digital Europe Programme administrative appropriations for external staff, and DEP Specific Objective 2 for operational expenditure
Searched Commission Decision of 24 January 2024 establishing the European AI Office, C/2024/1459, Article 8 (Official Journal PDF), snapshot 2026-09-15: Commission Decision C(2024) 390 Article 8 names only Union budget sources for the AI Office: DG CONNECT staff, Digital Europe Programme administrative appropriations for external staff, and DEP Specific Objective 2 for operational expenditure. confirmedhttps://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX%3A32024D01459
One hop away
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.