The operational governance architecture for educational AI
AIGN EOS —
Education
Operating System
For Trustworthy AI Decisions Affecting Children. The first operational, auditable and certifiable governance architecture for educational AI — globally applicable, legally grounded.
Three realities that apply right now — in every classroom.
No audit trail: When an AI decision disadvantages a student, no school without an EDR can document how that decision was reached — or defend it.
No liability basis: Without auditable decision records, there is no legal foundation to challenge or correct a flawed AI recommendation affecting a child.
No systemic protection: Algorithmic bias accumulates silently — without equity monitoring, structural discrimination becomes educational infrastructure.
Sections 1–5: architecture specification, normative synthesis, design science methodology.
Sections 2, 7 and 8: governance gap analysis, AIGN OS 2.0 integration, recommendations.
Section 6: Trust Label Levels 1–3, ASGR maturity, Education License.
No existing framework delivers
what schools actually need.
UNESCO provides orientation but no architecture. UNICEF provides principles but no protocol specification. OECD provides guardrails but no execution logic. The EU AI Act defines obligations but no implementation architecture for schools. AIGN EOS occupies precisely the space they leave open: the operational layer between norm and reality.
| Framework | What it delivers | What it leaves open | AIGN EOS response |
|---|---|---|---|
| UNESCO | Orientation, principles, policy framework | No implementation architecture; no artifact specification | Education Purpose Charter + D-CRIA Gate as implementation anchor |
| UNICEF v3.0 | Children’s rights principles; D-CRIA concept; protection requirements | No protocol specification; no auditability; no certification logic | EDR + Trust Label Levels 1–3 as auditable bridge from principle to evidence |
| OECD | Trustworthy AI principles; education guardrails; equity monitoring | No execution logic; no runtime layer; no lifecycle management | Policy-as-Code + Risk Engine as execution layer; ASGR as lifecycle monitor |
| EU AI Act | Legal obligations; high-risk classification; sanctions framework | No implementation architecture for schools; no sectoral runtime system | EDR as Art.-12-compliant logging system; ASGR as compliance evidence |
AIGN EOS as an Operating System:
Input → Processing → Output → Feedback.
Not a framework. Not a checklist. Not a policy document. AIGN EOS is a runtime layer that operates over educational AI systems — structuring decisions, recording them, verifying them and certifying them. Enforcement is achieved through Policy-as-Code gates, mandatory EDR writes, CRIA clearance and Trust Label gating.
Seven-layer governance architecture —
AIGN EOS as the Education Profile.
AIGN OS 2.0 is the horizontal governance architecture. AIGN EOS is its complete sectoral instantiation for education. Every EOS component is unambiguously assigned to one or more layers — fully traceable, no gaps.
Purpose & Scope D-CRIA Intake
Education Purpose Charter; use-case typology covering Admission / Assessment / Proctoring / Tutoring / Analytics. High-stakes classification at intake.
Risk & Impact Risk Engine
D-CRIA/CRIA as mandatory gate (Design→Pilot→Scale); Education Risk Register. Deployment blocked without CRIA clearance. Scales from pilot to full deployment.
Data & Privacy Policy-as-Code
Guardian Consent Flows; DPIA mapping; Data Minimization Defaults. UK Children’s Code-compliant. GDPR-aligned. Machine-readable rules enforced before every interaction.
Model & Evidence Equity Test Suite
Benefit-risk analysis before scaling. Bias/Equity Test Suite with subgroup-specific and continuous monitoring. Outcome evaluation as standard, not optional.
Transparency & Explainability EDR + Explainability UX
Multi-audience Explainability UX for 4 levels (Child · Parent · Teacher · Authority). Anti-anthropomorphizing rules. EDR disclosure fields for every decision output.
Oversight & Redress Human Override Record
Human-in/on-the-Loop Playbooks. Child-Rights Review Board. Appeal & Remedy Service with SLA and documented outcome. Every escalation path defined in advance.
Governance & Lifecycle Trust Label + ASGR
AIGN Education License; procurement clauses for EdTech vendors; Trust Label Levels 1–3; ASGR continuous maturity monitoring. Annual renewal and drift alerts.
The Education Decision Record:
If it isn’t recorded, it didn’t happen.
The EDR is the central intersection of all AIGN EOS components. It simultaneously serves as audit record, liability anchor, transparency instrument, redress foundation and certification evidence. A 12-field schema — mandatory, cryptographically signed, replay-capable.
Unique Decision Identifier
Links all subsequent records across the entire EDR chain. The spine of every auditable decision.
ISO-8601 Timestamp
Immutable record of when the decision occurred. Audit timeline and evidence preservation.
Classification
Admission / Assessment / Proctoring / Tutoring / Analytics. Determines applicable controls and high-stakes flag.
Input Data Categories
Categories of input data used (without raw data). Data protection compliance evidence at point of decision.
Model Identification
Unique model identification including version and training date. Enables reproducibility and drift detection.
Policy Verification Results
Pass/fail/overridden result of all Policy-as-Code verifications. Compliance evidence at the moment of every decision.
AI Output (Anonymized)
Actual AI output, anonymized/pseudonymized. The basis for contestation, review and appeal by affected parties.
Model Confidence
Model confidence measure for this decision. Uncertainty disclosure and escalation trigger for low-confidence outputs.
Human Oversight Record
Override: yes/no; substitution and reasoning if yes. Evidence of human oversight and liability anchoring for the school.
Contestation Link
Reference to complaint ID if contestation occurred. Traceability connecting remedy to original decision.
Longitudinal Outcome
Tracked consequence of the decision over time. Efficacy and fairness evidence for ongoing equity monitoring.
Cryptographic Hash
Cryptographic hash of the full EDR entry. Tamper-evident by design — integrity protection against manipulation.
The Control–Liability Logic.
AIGN EOS is built on a principle existing policy frameworks systematically omit: control generates accountability, and accountability requires a record. Without auditable decision records, there is no accountability. Without accountability, there is no incentive for governance.
Decision control
EDR records every AI decision: who, what, why, with what confidence, at what moment.
School can document and defend the decision before authorities or in court.
Process control
Policy-as-Code verifies compliance before every execution — machine-readable rules enforced at runtime.
Proof that institutional rules were active at the time of the decision.
Bias control
Equity Test Suite systematically detects disparate error rates across all relevant subgroups.
Documented bias evidence excludes silent accumulation of structural discrimination.
Override control
Human Override Record documents every human intervention with reasoning and timestamp.
Teacher and institutional responsibility is demonstrable — not merely asserted.
Remediation control
Appeal & Remedy Service with SLA and outcome protocol for children, parents and teachers.
Affected parties can verify whether their right to contest was effective or denied.
Education Trust Label —
Levels 1, 2 and 3.
The AIGN Education Trust Label makes governance maturity visible, comparable and procurement-relevant. Not a self-declared badge — an evidence-based certification level demonstrated through auditable artifacts, gated by ASGR maturity. Globally portable. Legally grounded.
Pilot-Ready
Operational
Assurable
Built for schools, universities
and EdTech providers worldwide.
AIGN EOS is designed for global portability. Trust Label Level 1 requires no advanced digital infrastructure — making it applicable from primary schools in rural contexts to research universities in regulated markets.
Your school uses AI.
Can you prove it’s safe?
From adaptive learning tools to proctoring systems and admission algorithms — every AI touching a student’s path requires governance. AIGN EOS starts at Level 1: structured documentation, a basic CRIA process and foundational training. No advanced infrastructure required.
Lead the field.
Certify your governance.
Universities are simultaneously providers, deployers and researchers of AI. AIGN EOS provides the institutional architecture to govern all three roles — from student-facing systems to research AI and administrative automation — with a single certifiable framework.
Build products that institutions can actually procure.
From 2027, EU AI Act Art. 12 applies to high-risk educational AI. Without a demonstrable EDR export interface and audit-ready logging, your product lacks the compliance foundation required for institutional procurement. AIGN EOS defines exactly what’s required — and how to deliver it.
Every school. Every country. Every child deserves accountable AI.
AIGN EOS is designed for global portability — from European universities navigating the EU AI Act to schools in Sub-Saharan Africa, South Asia and Latin America deploying AI-driven platforms through donor programs. Trust Label Level 1 requires no advanced infrastructure. Begin now.
Four phases from first assessment
to certified, auditable assurance.
Discovery
Typify use cases; high-stakes classification; D-CRIA intake; data flow mapping; stakeholder plan including child participation; ASGR baseline assessment.
→ ASGR Level 1 evidenceBuild
Implement EDR standard; establish RMF-based lifecycle; minimal controls (logging, oversight, disclosure); AIGN Academy foundational training; pilot evaluation.
→ Trust Label Level 1Scale
Activate Evidence-Before-Scale Gate; conclude Education License agreements; launch equity dashboard; full CRIA report; ASGR Level 2 evidence package.
→ Trust Label Level 2Assure
Third-party audit; ISO/IEC-42001 alignment; independent outcome evaluation; full ASGR Level 3 maturity assessment. Certifiable governance achieved.
→ Trust Label Level 3What must change — for governments,
EdTech providers and schools.
Policy & statutory obligations
Product & compliance requirements
Operational governance requirements
What distinguishes AIGN EOS from every other approach.
Five operationalized propositions — not principles, not guidelines, not intentions. Each is enforced at runtime, auditable and certifiable. AIGN EOS provides the operational architecture. Not another guideline.
Educational AI systems that lack an EDR cannot systematically meet the logging and traceability requirements of EU AI Act Art. 12. No decision without a record.
Educational AI deployed without CRIA significantly increases structural bias risks and undermines non-discrimination obligations. No deployment without impact assessment.
Educational AI without a Human Override Record does not operationalize EU AI Act Art. 14 human oversight. Responsibility must be demonstrable, not merely asserted.
Educational AI without Trust Label Level 2 lacks a structured evidence pathway to demonstrate conformity with high-risk AI requirements from 2027.
AIGN EOS provides the operational architecture. Not another guideline. Not a framework. An operating system — enforceable, auditable, certifiable.
Your school. Your students.
Your accountability.
The EU AI Act high-risk obligations for educational AI apply from 2027. AIGN EOS provides the architecture to meet them — auditable, certifiable, globally applicable. Start in days.