Skip to content
AIGP

Jurisdictional AI Governance — Implementation Guide

Jurisdictional AI Governance — Implementation Guide

PRIVATE AND PROPRIETARY — NOT A PUBLIC RFC. Owned by Kanjani AI Research & Causum. See NOTICE.md.

How to declare and operate under jurisdiction-specific AI governance contexts using AIGP v4.0.

Overview

AIGP v4.0 supports five governance layers, evaluated in strict priority order:

Priority Context RFC Use Case
1 (highest) Humanitarian (IHL) RFC-027 Conflict zones, humanitarian operations
2 EU AI Act RFC-028 EU-deployed AI, EU data subjects
3 AU Continental Strategy RFC-029 African nations, AU member states
4 Japan AI Promotion Act RFC-030 Japan-deployed AI, Japanese data subjects
5 Universal Humanity RFC-031 Baseline governance for all AI systems

Quick Start

Python

from aigp_client import AigpClient, JapaneseContext, AfricanContext


client = AigpClient("https://gov.example.com", "my-app", "hmac-secret")


# Declare jurisdiction — pick your context
client.declare_japanese_context(JapaneseContext(
    active=True,
    governance_framework="JAPAN_AI_PROMOTION_ACT",
    innovation_priority=True,
    sector="HEALTHCARE",
    hiroshima_principles=["TRUSTWORTHY_AI", "RISK_BASED_APPROACH"],
))


# All subsequent checks are evaluated against Japan rules
decision = await client.check("diagnosis-assist", "claude-4")

TypeScript

const client = new AigpClient("https://gov.example.com", "my-app", "hmac-secret");


client.declareJapaneseContext({
  active: true,
  governanceFramework: "JAPAN_AI_PROMOTION_ACT",
  innovationPriority: true,
  sector: "HEALTHCARE",
});


const decision = await client.check("diagnosis-assist", "claude-4");

Go

client := aigp.NewClient("https://gov.example.com", "my-app", "hmac-secret")
client.DeclareJapaneseContext(ctx, aigp.JapaneseContext{
    Active:              true,
    GovernanceFramework: "JAPAN_AI_PROMOTION_ACT",
    InnovationPriority:  true,
    Sector:             "HEALTHCARE",
})

Evaluation Order

When multiple contexts are declared, rules are evaluated in priority order. The first DENY stops execution:

REQUEST arrives
  → IHL non-derogable rules (RFC-027)
  → EU AI Act rules (RFC-028)
  → AU Strategy rules (RFC-029)
  → Japan Act rules (RFC-030)
  → Universal rules (RFC-031)
  → Configurable policies (scope envelopes, budgets)
  → ALLOW (if nothing denied)

Key principle: Higher-priority frameworks cannot be overridden by lower ones. IHL rules are absolute.

Multi-Jurisdiction Scenarios

Cross-border AI deployment

An AI system deployed across multiple jurisdictions should declare ALL applicable contexts:

# EU-headquartered company deploying in Kenya
client.declare_regulatory_context(RegulatoryContext(
    active=True, framework="EU_AI_ACT",
    risk_classification="HIGH_RISK",
    deployer_jurisdiction="KE",
    provider_jurisdiction="DE",
))
client.declare_african_context(AfricanContext(
    active=True,
    governance_framework="AU_CONTINENTAL_AI_STRATEGY",
    member_state="KE", rec="EAC",
    ubuntu_principle=True,
))

Both rule sets apply. EU transparency requirements AND AU data sovereignty requirements must be met.

Conflict zones

When AI operates in an active conflict area, IHL context takes absolute precedence:

client.declare_humanitarian_context(HumanitarianContext(
    active=True,
    legal_regime="IHL_GCIV",
    conflict_classification="NIAC",
    operating_authority="ICRC",
    protected_person_categories=["CIVILIAN", "MEDICAL_PERSONNEL"],
))

IHL non-derogable rules cannot be overridden by ANY other context.

Japan-Specific Guidance

Innovation-First Principle

Japan’s Act frames governance as enabling innovation rather than restricting it. AIGP implements this via:

  • JP-001 (INNOVATION_ENABLEMENT): DENY only triggers if a restriction cannot demonstrate proportional justification
  • JP-004 (SAFETY_RELIABILITY): Safety measures are required for high-risk sectors but expressed as positive obligations (what you MUST do) not prohibitions

Transparency Levels

Level When Required Obligation
STANDARD All AI systems Basic disclosure that AI is in use
ENHANCED Public-facing, high-impact Full disclosure of AI decision factors
RESEARCH R&D systems Methodology documentation

Sectors

Declare the deployment sector to activate sector-specific rules: HEALTHCARE, EDUCATION, MANUFACTURING, PUBLIC_SERVICES, FINANCE, AGRICULTURE, DEFENSE

Universal Context (RFC-031)

The universal context activates baseline governance that applies regardless of jurisdiction:

# Activate universal governance (all 10 principles)
client.declare_universal_context({
    "active": True,
    "principles": "ALL",
    "accountability_contact": "governance@example.com",
})

Universal rules (UH-001 through UH-010) provide minimum governance for AI systems that don’t fall under any specific jurisdiction.

Server Configuration

The governance server automatically evaluates all declared contexts. No additional server configuration is needed — contexts are declared per-app via the SDK.

Verifying Context Declaration

Terminal window
curl -X POST https://gov.example.com/api/v1/context/declare \
  -H "Content-Type: application/json" \
  -H "X-AIGP-App-Id: my-app" \
  -d '{"app_id":"my-app","japanese_context":{"active":true,"governance_framework":"JAPAN_AI_PROMOTION_ACT"}}'

Checking Active Contexts

Active contexts are returned in CHECK responses when relevant rules trigger:

{
  "decision": "ALLOW_WITH_CONSTRAINTS",
  "constraints": {
    "jp_transparency": "ENHANCED",
    "jp_safety_required": true
  }
}

Evidence and Audit

All jurisdictional governance decisions are logged with:

  • Which context triggered the decision
  • Which rule ID (e.g., JP-002, AU-004, EUAI-003)
  • Whether the rule is non-derogable
  • Timestamp and request metadata

This creates a compliance evidence trail for auditors in any jurisdiction.

Further Reading