A governed methodology for evidence-backed AI governance opinions.
An AIGR rating is a point-in-time governance opinion within a defined entity, system, lifecycle, jurisdiction and evidence period. The public methodology describes the rating architecture, evidence standard, review controls and rating actions while keeping proprietary weights, thresholds and calibration logic controlled.
Defensibility comes from disciplined governance of the rating.
The rating process is designed around explicit scope, evidence traceability, controlled judgment, separate issuance authority and ongoing monitoring. The objective is reproducibility from the methodology version, assessment record and retained decision history.
Scope
Define the subject of the opinion before assessment begins.
Evidence
Anchor material conclusions to reviewable records.
Judgment
Record exceptions, findings and material interpretation.
Issuance
Separate final rating approval from routine assessment activity.
Monitoring
Track material change, incidents, remediation and rating triggers.
The analytical score and the published rating designation serve different purposes.
Separating diagnostic analysis from the categorical outcome preserves analytical depth without presenting the rating as a percentage, probability or credit score.
Governance Score
A 0–100 analytical measure supports diagnostics, remediation prioritization and longitudinal trend analysis. Calculation logic remains controlled.
AIGR designation
A categorical designation from AIGR-100 through AIGR-40, plus AIGR-NR, provides the executive-level governance opinion.
Eight domains create the public enterprise governance view.
Sector and jurisdiction overlays adjust materiality and evidence expectations while preserving a consistent core architecture.
A controlled decision process from intake through monitoring.
Intake
Define organization, AI system, use case, sector and perimeter.
Classify
Set risk context, materiality and evidence expectations.
Assess
Evaluate controls, evidence, findings and deficiencies.
Review
Challenge evidence quality, interpretation, exceptions and conflicts.
Rate
Approve the designation, rationale, conditions and decision record.
Monitor
Track validity, change triggers, incidents and re-review.
Seven categorical designations, plus Not Rated.
The designation summarizes governance maturity and evidentiary confidence inside the stated perimeter.
AIGR-100 through AIGR-40 are categorical designations. They are not percentages, probabilities, credit ratings or guarantees. The separate Governance Score supports analysis but is not itself the published rating designation.
Five parameters define the rated perimeter.
Four tests determine whether evidence can support a conclusion.
A policy is not sufficient merely because it exists. Evidence must be fit for the requirement and capable of supporting the operating conclusion being reached.
Relevance
The artifact addresses the control requirement directly.
Currency
The artifact reflects the system and evidence period in scope.
Traceability
Owner, date, source and evidence lineage remain attributable.
Sufficiency
The artifact demonstrates operating practice rather than design intent alone.
Verified
Current, attributable and sufficient evidence supports the conclusion.
Qualified / partial
Relevant evidence exists but limitations reduce confidence.
Self-attested
Management assertion is recorded without sufficient reviewable support.
Absent / blocker
Required evidence is missing, rejected or a material gate prevents reliance.
Assessment, remediation support and rating issuance are separated.
The governance model is designed so commercial or advisory activity does not determine the published designation.
Assessment
Collects and tests evidence, records findings and establishes control conclusions.
Remediation support
Supports closure of governance gaps and remains separate from final rating approval.
Rating issuance
Challenges the record and approves, modifies or declines the designation.
Potential conflicts are identified at intake and retained in the rating record. Where conflicts cannot be managed through appropriate separation or other controls, the engagement can be declined or the outcome designated AIGR-NR.
An issued rating remains subject to defined actions.
Affirmation
Current evidence continues to support the designation.
Upgrade / downgrade
Material change in governance capability or evidence changes the designation.
Conditional
A material gate or unresolved condition constrains the outcome.
Suspension
The opinion is paused while a material incident, change or evidence issue is investigated.
Withdrawal
The rated scope no longer applies or continued use of the rating is not supportable.
Not Rated
Scope, evidence, independence or assessment conditions do not support an opinion.
Material factual or methodological issues can be challenged through a recorded process.
Grounds can include factual error, material evidence not considered, misapplication of the methodology or material scope misstatement. The determination is retained whether or not the designation changes.
The framework itself is controlled and versioned.
Versioning
Issued ratings identify the methodology version and effective period applied.
Change control
Material changes to criteria, gates or scale are documented prospectively.
Calibration
Decision consistency is evaluated across assessments and reviewers.
Confidentiality
Client evidence remains tenant-segregated and is not reproduced in public outputs without authorization.
Benchmarks
Cohorts are constructed subject to consent, confidentiality and sufficient comparability.
Records
Evidence, findings, review challenge, approvals and rating actions are retained for reconstruction and appeal.
One governance architecture can support multiple sector and regulatory contexts.
Mappings provide traceability and readiness support. They do not convert an AIGR rating into a certification, statutory conformity assessment or legal conclusion.
Interpretation remains bounded by the scope and evidence.
An opinion on governance maturity and evidentiary confidence within the defined assessment perimeter.
A point-in-time decision-support signal subject to monitoring and defined rating actions.
A certification, accreditation, legal opinion, audit opinion or regulatory approval.
A percentage, probability, credit rating, financial recommendation or safety guarantee.