Defined reviewer role
001 asks whether a real reviewer is identified and capable of carrying the role.
Claim-to-source mapNIST GOVERN 2.1 + GOVERN 3.2; ICO reviewer responsibility/qualification guidance
Why the tool asks what it asks.
001 translates recurring human-oversight concepts into a practical diagnostic. This record distinguishes direct source support, cross-source synthesis, and product heuristics instead of presenting the whole framework as if one authority prescribed it.
Enough information / evidence
001 asks whether the reviewer can see enough context to interpret and challenge the AI-supported decision.
Claim-to-source mapNIST MAP 2.2; EU AI Act Arts. 13–14; ICO meaningful-review guidance
Timing before consequence
001 treats timing as a design condition: a safeguard cannot affect an outcome after the relevant intervention window has closed.
Claim-to-source mapEU AI Act Art. 14 intervention/override/stop concepts; NIST MAP 3.5; ICO appropriate review stages
Authority to disagree / override / stop
001 distinguishes observation/advice from operative power to change the decision or trigger a fallback/escalation.
Claim-to-source mapEU AI Act Art. 14; ICO meaningful influence/authority; NIST human-oversight roles
Escalation / fallback
001 recommends a path when the reviewer cannot safely resolve the issue themselves.
Claim-to-source mapICO fallback/manual options; NIST risk-response and oversight processes
Record of review / override
001 treats a review record as evidence that review occurred and as input to later evaluation.
Claim-to-source mapICO override logs; NIST documentation and monitoring functions
Composite score / priority ordering
The score is a design aid created for this tool. It is not a legal, regulatory, safety, or compliance score.
Claim-to-source mapNo source mandates the product score used by 001.
Exact provenance used for Build 001
NIST — AI Risk Management Framework 1.0 — Core + Human-AI Interaction
Published/version: 2023-01-26
Relevant sections: GOVERN 2.1; GOVERN 3.2; MAP 2.2; MAP 3.4–3.5; MEASURE 1.2–1.3; Appendix C
Supports: defined roles and responsibilities; differentiated human-AI oversight roles; sufficient information for human decisions; operator proficiency; defined/assessed oversight; evaluation of control effectiveness; independent/domain review.
Current-status note: AI RMF 1.0 remains published; NIST states a revision is in progress.
European Union — Regulation (EU) 2024/1689 — Article 14, Human oversight
Published/version: 2024-07-12
Relevant sections: Article 14; related deployer-information requirements in Article 13
Supports: effective oversight of high-risk AI; appropriate human-machine interfaces; understanding capabilities/limitations; awareness of automation bias; correct interpretation; ability to disregard/override/reverse output or interrupt the system where provided by Article 14.
Current-status note: Binding EU regulation; applicability depends on system, role, jurisdiction, and dates of application.
UK Information Commissioner’s Office — Artificial intelligence audit toolkit — Human review
Published/version: current web guidance
Relevant sections: Human review control measures, expectations, and options
Supports: meaningful review; reviewer qualifications/training; manageable workload/time; independence; ability to influence decisions; logging overrides/reasons; fallback/manual paths; standardized review procedures; testing review effectiveness.
Current-status note: ICO explicitly says this guidance is under review following the Data (Use and Access) Act and may change.
Important: these sources do not collectively certify 001 as a compliance instrument. Applicability is context- and jurisdiction-dependent, and the tool intentionally labels its composite score and prioritization logic as product heuristics.