The 52-Article Charter · 19 of 52 · full text

Article 19: International Regulatory Integration

Published from the canonical CSOAI Partnership Charter (effective 15 January 2026). Full text below.

Version: 1.0 Effective Date: January 15, 2026, 09:00 GMT Status: Integration Article - Regulatory Harmonization

PREAMBLE

This Article establishes how CSOAI integrates with existing and emerging international AI regulatory frameworks. CSOAI does not replace national regulations—it complements and harmonizes them. One Charter, many jurisdictions, unified safety.

Core Principle: Regulatory interoperability through mutual recognition and coordinated standards.

19.1 EU AI ACT INTEGRATION

19.1.1 Mutual Recognition Framework

EU AI Act (Regulation 2024/1689):

CSOAI License Satisfies EU Requirements:

For High-Risk AI Systems (EU AI Act Article 6):

CSOAI High-Risk or Critical license = EU conformity presumption
CSOAI safety case = EU technical documentation (Annex IV)
CSOAI audits = EU conformity assessment (Annex VII)
Byzantine Council monitoring = EU post-market monitoring (Article 61)

Mapping CSOAI to EU Requirements:

| EU AI Act Requirement | CSOAI Charter Equivalent |
|----------------------|-------------------------|
| Risk Management System (Art. 9) | Safety Case (Article 2) |
| Data Governance (Art. 10) | Training Data Review (Article 15.3.3) |
| Technical Documentation (Art. 11) | License Application (Article 10.3) |
| Record-Keeping (Art. 12) | Byzantine Council Logs (Article 11.3.1) |
| Transparency (Art. 13) | Public Watchdog (Article 13) |
| Human Oversight (Art. 14) | Human Council (Article 12) |
| Accuracy, Robustness (Art. 15) | Provable Safety (Article 2) |
| Conformity Assessment (Art. 43) | Compliance Assessment (Article 15) |
| Post-Market Monitoring (Art. 61) | Continuous Monitoring (Article 11) |
| Serious Incident Reporting (Art. 62) | Incident Reporting (Article 13.4) |

19.1.2 EU Notified Body Status

CSOAI Application for EU Notified Body Designation:

Requirements to Become Notified Body:

Independence and impartiality ✅
Technical competence in AI ✅
Documented procedures ✅ (this Charter)
Quality management system ✅
Professional liability insurance ✅

Application Process:

Submit to EU Commission (Q1 2026)
Demonstrate compliance with Regulation 765/2008
Peer assessment by existing notified bodies
Designation by Member State (UK initially, then others)
Notification to EU Commission
Listed in NANDO database

Benefits of Notified Body Status:

CSOAI conformity assessments legally binding in EU
Single assessment = EU + global compliance
Revenue from EU conformity services
Influence over EU AI safety standards

Timeline: Apply Q1 2026, designation by Q4 2026

19.1.3 Prohibited AI Systems (EU Article 5)

EU Bans Certain AI Applications:

Subliminal manipulation causing harm
Exploiting vulnerabilities (age, disability)
Social scoring by public authorities
Real-time biometric identification (law enforcement, limited exceptions)
Biometric categorization based on sensitive attributes
Emotion recognition (workplace, education)
Predictive policing based solely on profiling

CSOAI Position:

Alignment: CSOAI prohibits same systems (Constitutional AI Article 5 prohibits manipulation, discrimination)

Additional Restrictions: CSOAI goes further:

No military AI without human-in-loop (Article 16.5.1)
No AI concealing consciousness emergence
No AI designed to deceive users systematically

License Denials: Applications for prohibited uses automatically denied

19.1.4 CE Marking and Declaration of Conformity

CSOAI-Licensed AI in EU:

Process:

Obtain CSOAI High-Risk license
CSOAI issues EU conformity certificate
Manufacturer affixes CE marking
Submits Declaration of Conformity to EU database
Can deploy in all EU Member States

CSOAI Declaration Template: ``` DECLARATION OF CONFORMITY

Manufacturer: [Company Name] AI System: [System Name] CSOAI License: [License Number]

We declare under our sole responsibility that the AI system identified above is in conformity with:

EU AI Act (Regulation 2024/1689)
CSOAI Partnership Charter Articles 1-52

The conformity assessment was performed by: Council for the Safety of AI (CSOAI) Notified Body No. [XXXX]

Conformity Certificate: [Certificate Number] Date: [Date]

Authorized Signatory: [Name, Title] ```

Surveillance: EU market surveillance authorities can verify, CSOAI cooperates

19.2 NIST AI RISK MANAGEMENT FRAMEWORK INTEGRATION

19.2.1 NIST AI RMF Mapping

NIST AI RMF 1.0 (January 2023):

Four Core Functions:

1. GOVERN:

NIST: Organizational policies, culture, risk tolerance
CSOAI: Articles 9-12 (governance structure, Human Council)
Integration: CSOAI membership = NIST governance framework adoption

2. MAP:

NIST: Context, categorization, expectations
CSOAI: Article 10 (licensing, risk tiers)
Integration: CSOAI risk categorization satisfies NIST mapping

3. MEASURE:

NIST: Assessment, benchmarking, metrics
CSOAI: Article 15 (compliance assessment, audits)
Integration: CSOAI audits provide NIST measurement evidence

4. MANAGE:

NIST: Prioritize, respond, resources
CSOAI: Article 17 (enforcement), Article 8 (prosperity resources)
Integration: CSOAI enforcement = NIST risk management

NIST Trustworthiness Characteristics:

| NIST Characteristic | CSOAI Implementation |
|-------------------|---------------------|
| Valid & Reliable | Provable Safety (Article 2) |
| Safe | Byzantine Council monitoring (Article 11) |
| Secure & Resilient | Cybersecurity requirements (Article 27) |
| Accountable & Transparent | Public Watchdog (Article 13) |
| Explainable & Interpretable | Mechanistic Interpretability (Article 2.2) |
| Privacy-Enhanced | Data protection standards (Article 29) |
| Fair (bias managed) | Constitutional AI fairness (Article 5) |

19.2.2 NIST Voluntary Commitment

CSOAI as NIST Framework Implementer:

Commitment to NIST:

Adopt AI RMF as baseline standard
Require all licensees to implement RMF
Audit for RMF compliance
Report NIST conformance annually
Participate in NIST community of practice

Benefits:

US federal recognition (government AI requires NIST compliance)
Industry standard adoption (NIST widely respected)
Harmonization with US regulatory approach
Influence on NIST evolution

Letter to NIST Director: ``` Dr. Laurie E. Locascio, Director National Institute of Standards and Technology

Re: CSOAI Commitment to NIST AI RMF

Dear Dr. Locascio,

The Council for the Safety of AI (CSOAI) commits to full adoption of NIST AI RMF 1.0 as a foundational element of our Charter.

We will require all CSOAI-licensed AI systems to demonstrate NIST RMF compliance as part of the licensing process.

We request formal recognition as a NIST AI RMF implementing organization and look forward to collaboration.

Sincerely, Nicholas Tonna Executive Director, CSOAI ```

19.2.3 NIST Special Publication 800 Series

Integration with NIST Cybersecurity Standards:

NIST SP 800-53 (Security Controls):

AI systems must implement appropriate controls
CSOAI requires 800-53 compliance for Medium+ risk tiers
Annual 800-53 assessment part of audit

NIST SP 800-160 (Systems Security Engineering):

Lifecycle security for AI systems
CSOAI adopts systems engineering approach
Security-by-design requirement

NIST SP 800-218 (Secure Software Development):

AI software development practices
Code review, testing, vulnerability management
CSOAI audits verify compliance

19.3 ISO/IEC AI STANDARDS INTEGRATION

19.3.1 ISO/IEC 42001 (AI Management System)

ISO/IEC 42001:2023 - Artificial Intelligence Management System:

CSOAI License Requires ISO 42001:

All Medium+ risk AI must achieve ISO 42001 certification
CSOAI auditors verify ISO 42001 compliance
Annual recertification required

ISO 42001 Elements:

| ISO 42001 Clause | CSOAI Implementation |
|-----------------|---------------------|
| 4. Context | License application (Article 10.3) |
| 5. Leadership | Executive accountability (Article 17.5.1) |
| 6. Planning | Safety case (Article 2.4) |
| 7. Support | Staff training, resources |
| 8. Operation | Deployment monitoring (Article 11) |
| 9. Performance | Compliance assessment (Article 15) |
| 10. Improvement | Corrective action (Article 17.7) |

Certification Bodies:

CSOAI collaborates with accredited ISO 42001 certification bodies
Mutual recognition: ISO 42001 + CSOAI license = comprehensive compliance

19.3.2 ISO/IEC 23894 (AI Risk Management)

ISO/IEC 23894:2023 - Guidance on Risk Management:

CSOAI Risk Framework Based on ISO 23894:

Risk assessment methodology
Risk treatment options
Continuous risk monitoring
Integration with NIST AI RMF (they're aligned)

CSOAI Adds:

Mandatory formal verification (beyond ISO guidance)
Byzantine Council continuous monitoring (automated risk detection)
Human Council oversight (governance layer)

19.3.3 ISO/IEC 22989 (AI Concepts and Terminology)

Standardized Terminology:

CSOAI adopts ISO 22989 definitions:

Artificial Intelligence
Machine Learning
Neural Network
Training, Testing, Validation
Bias, Fairness, Explainability

Ensures: Consistent language across regulations, industries, jurisdictions

19.3.4 ISO/IEC 38507 (Governance of IT - AI Extension)

Governance Framework:

CSOAI governance structure (Articles 9-12) aligns with ISO 38507:

Board oversight
Executive accountability
Stakeholder engagement
Performance measurement
Conformance assessment

19.4 ADDITIONAL REGULATORY FRAMEWORKS

19.4.1 OECD AI Principles

OECD Principles on AI (2019):

Inclusive growth, sustainable development, well-being

- CSOAI: Prosperity Covenant (Article 8)

Human-centered values and fairness

- CSOAI: Maternal Covenant (Article 1), Constitutional AI (Article 5)

Transparency and explainability

- CSOAI: Public Watchdog (Article 13), Interpretability (Article 2.2)

Robustness, security, safety

- CSOAI: Provable Safety (Article 2), Byzantine Council (Article 11)

Accountability

- CSOAI: Enforcement (Article 17), Appeals (Article 18)

CSOAI Recognition by OECD:

Apply for OECD observer status
Participate in OECD AI expert groups
Align with OECD recommendations
Report OECD principle compliance

19.4.2 UNESCO AI Ethics Recommendation

UNESCO Recommendation on Ethics of AI (2021):

Four Values:

Human rights and human dignity → CSOAI Article 1, 5
Living in peaceful, just, interconnected societies → CSOAI Article 7, 8
Environmental and ecosystem flourishing → CSOAI sustainability (Article 31)
Diversity and inclusiveness → CSOAI Article 14

Ten Principles:

Proportionality, do no harm, safety, fairness, etc.
All addressed in CSOAI Charter

CSOAI Commitment:

Implement UNESCO recommendation
Report compliance to UNESCO
Participate in UNESCO AI ethics dialogues

19.4.3 UK AI Regulation (Pro-Innovation Approach)

UK Government AI White Paper (March 2023):

Five Principles:

Safety, security, robustness → CSOAI Articles 2, 11
Appropriate transparency → CSOAI Article 13
Fairness → CSOAI Article 5
Accountability → CSOAI Article 17
Contestability, redress → CSOAI Article 18

UK Context-Specific Regulators:

FCA (finance), MHRA (medicine), Ofcom (communications), etc.
CSOAI collaborates with sector regulators
Shared jurisdiction: CSOAI (AI safety), sectoral (domain expertise)

CSOAI as UK Success Story:

Registered in England (UK entity)
Demonstrates UK AI leadership
Pro-innovation (enables safe deployment)
Global reach (UK soft power)

19.4.4 Singapore Model AI Governance Framework

Singapore PDPC Model Framework (2020, updated 2022):

CSOAI Alignment:

Internal governance structures ✅ (Article 9)
Human oversight ✅ (Article 12)
Operations management ✅ (Article 11)
Stakeholder interaction ✅ (Article 14)

Singapore Collaboration:

CSOAI Asia-Pacific office in Singapore
Work with PDPC on harmonization
Singapore as AI governance hub

19.4.5 China AI Regulations

Multiple Chinese Regulatory Measures:

Algorithm Recommendation Regulations (2022)
Deep Synthesis Regulations (2023)
Generative AI Measures (2023)

CSOAI Approach:

Monitor Chinese regulations closely
Seek common ground (safety, fairness)
Respect sovereignty while maintaining Charter principles
Encourage Chinese companies to adopt CSOAI standards voluntarily

Challenge: Chinese regulatory environment differs significantly

State control vs. independent oversight
Different values framework
National security primacy

Strategy:

Engage pragmatically
Focus on technical safety (common interest)
Respect differences on governance

19.5 SECTOR-SPECIFIC REGULATIONS

19.5.1 Medical Device Regulations (FDA, EMA, MHRA)

AI as Medical Device:

FDA (US):

Software as Medical Device (SaMD)
AI/ML-based SaMD Action Plan
Predetermined Change Control Plan

CSOAI Integration:

Medical AI requires both CSOAI license AND FDA approval
CSOAI safety case supports FDA 510(k) or PMA submission
Byzantine Council monitoring complements FDA post-market surveillance

EMA (EU):

Medical Device Regulation (MDR 2017/745)
In Vitro Diagnostic Regulation (IVDR 2017/746)

CSOAI Integration:

CSOAI conformity assessment can satisfy MDR/IVDR for AI components
Collaborate with EU Notified Bodies for medical devices

MHRA (UK):

Software and AI as Medical Device Change Programme
MHRA recognizes CSOAI for AI safety assessment

19.5.2 Financial Services Regulations

AI in Finance:

FCA (UK), SEC (US), ESMA (EU):

Model risk management
Algorithmic trading oversight
Consumer protection (unfair outcomes)

CSOAI Requirements for Financial AI:

Financial AI = High Risk tier (Article 10)
Additional requirements:

- Fair lending analysis (bias testing) - Explainability for credit decisions - Market manipulation prevention - Systemic risk assessment

Coordination:

CSOAI handles AI safety
Financial regulators handle financial soundness
Joint oversight for maximum effectiveness

19.5.3 Autonomous Vehicle Regulations

Transportation Safety:

NHTSA (US), UNECE (Europe), MLIT (Japan):

Vehicle safety standards
Autonomous driving regulations
Type approval processes

CSOAI Role:

AI system safety certification
Ethical decision-making framework (trolley problem)
Continuous monitoring

Vehicle Manufacturer Role:

Mechanical safety (existing standards)
Integration with CSOAI-certified AI

Example: Tesla Autopilot:

NHTSA: Vehicle safety, crash investigation
CSOAI: AI decision-making, value alignment, Byzantine Council monitoring
Complementary oversight

19.5.4 Employment and Labor Law

AI in Hiring/HR:

EEOC (US), EHRC (UK), EU Anti-Discrimination Directives:

Prohibited discrimination
Disparate impact analysis
Right to explanation

CSOAI Requirements:

Hiring AI = Medium-High Risk
Mandatory bias testing (protected characteristics)
Explainability for adverse decisions
Human review of final decisions
Right to appeal (Article 18)

19.6 MUTUAL RECOGNITION AGREEMENTS

19.6.1 Framework for Mutual Recognition

CSOAI Seeks Agreements With:

Major Jurisdictions:

European Union (via Notified Body status)
United States (via NIST framework alignment)
United Kingdom (home jurisdiction)
Singapore (Asia-Pacific hub)
Canada, Australia, Japan, South Korea

Agreement Template: ``` MUTUAL RECOGNITION AGREEMENT Between: [Jurisdiction] and CSOAI

Article 1: Recognition [Jurisdiction] recognizes CSOAI licensing as meeting [Jurisdiction's] AI safety requirements for [specified risk categories].

Article 2: Scope Applies to: [High-Risk AI systems] OR [All AI systems] Excludes: [National security applications, sector-specific where needed]

Article 3: Information Sharing Parties commit to sharing:

Incident reports
Best practices
Research findings
Regulatory updates

Article 4: Cooperation Joint enforcement actions where appropriate Coordinated responses to cross-border violations

Article 5: Review Agreement reviewed annually, updated as needed ```

19.6.2 Benefits of Mutual Recognition

For Companies:

Single license, multiple jurisdictions
Reduced compliance burden
Faster time to market
Lower costs

For Regulators:

Shared resources (can't all build AI expertise equally)
Coordinated oversight (AI is global, regulation must be too)
Race to the top (harmonized high standards, not race to bottom)

For Public:

Consistent protection globally
No regulatory gaps
Transparent, coordinated enforcement

19.6.3 Jurisdictional Opt-In

Countries Can Choose:

Full Recognition:

CSOAI license sufficient for all AI deployment
Country relies entirely on CSOAI
Country participates in CSOAI governance

Partial Recognition:

CSOAI license required but not sufficient
Additional national requirements
Coordination on enforcement

No Recognition:

Country develops own framework
CSOAI respects sovereignty
Companies must comply with both

Reality: Most likely outcome is Partial Recognition (CSOAI + national)

19.7 INTERNATIONAL TREATY FRAMEWORK

19.7.1 UN Convention on AI Safety (Proposed)

CSOAI Advocates For:

International Treaty on AI Governance:

Similar to: IAEA (nuclear), OPCW (chemical weapons), UNFCCC (climate)
Purpose: Binding international commitments on AI safety
Scope: AGI/ASI safety, existential risk prevention, prosperity sharing

Draft Treaty Elements:

Article I: Obligations

Parties commit to safe AI development
Licensing of high-risk AI
Prosperity sharing mechanisms
Emergency cooperation protocols

Article II: CSOAI Recognition

Treaty recognizes CSOAI as implementation mechanism
CSOAI Charter incorporated by reference
Countries can adopt CSOAI directly or create compatible frameworks

Article III: Verification

Inspections of AI facilities
Reporting requirements
Byzantine Council access to monitor compliance

Article IV: Enforcement

Treaty violations referred to International Court of Justice
Sanctions for non-compliance
Emergency intervention for existential threats

Timeline:

Draft treaty: 2026
Negotiation: 2027-2028
Adoption: UN General Assembly 2029
Entry into force: 2030

19.7.2 G20 and Multilateral Forums

G20 AI Principles:

CSOAI presents Charter to G20 Digital Economy Working Group
Seek G20 endorsement
Influence G20 ministerial declarations

Other Forums:

World Economic Forum (Davos)
OECD AI Policy Observatory
UN AI Advisory Body
ITU AI for Good Summit

Strategy:

Build consensus through multiple channels
Demonstrate CSOAI as credible solution
Pragmatic implementation path

19.8 COMPLIANCE DEMONSTRATION

19.8.1 Regulatory Crosswalk Documents

CSOAI Publishes Crosswalk Guides:

"How CSOAI License Satisfies EU AI Act Requirements"

Line-by-line mapping
Evidence packages
Templates for compliance documentation

"CSOAI and NIST AI RMF: Integration Guide"

RMF function mapping
Evidence collection
Audit procedures

Similar Guides for:

ISO standards
Sector-specific regulations
National frameworks

Available:

Free download from csoai.org/regulatory
Updated as regulations evolve
Translated into major languages

19.8.2 Joint Audits

Where Possible:

Coordinated Regulatory Inspection:

CSOAI auditor + EU Notified Body + NIST assessor
Single audit, multiple compliance checks
Shared evidence base
Efficient for companies

Example: Medical AI device deployed in US and EU:

Day 1: CSOAI safety assessment
Day 2: FDA SaMD review
Day 3: EU MDR conformity
Single comprehensive audit, three certifications

19.8.3 Regulatory Reporting

CSOAI Reports to Regulators:

Annual Regulatory Compliance Report:

Submit to: EU Commission, NIST, OECD, WHO, etc.
Content:

- Number of licenses issued - Compliance statistics - Incident summaries - Enforcement actions - Research findings

Ad Hoc Reporting:

Serious incidents (immediate notification)
Emerging risks (as identified)
Policy recommendations (evidence-based)

19.9 HARMONIZATION CHALLENGES

19.9.1 Regulatory Divergence

Challenge: Different jurisdictions, different approaches

Examples:

EU: Comprehensive ex-ante regulation (AI Act)
US: Sector-specific, principles-based, light touch
China: State control, national security focus
Singapore: Pro-innovation, soft law

CSOAI Strategy:

Find common denominators (safety, fairness, transparency)
Flexible implementation (adapt to local context)
Core principles non-negotiable (Charter articles 1-8)
Governance adjustable (articles 9+)

19.9.2 Sovereignty Concerns

Challenge: Countries don't want to outsource regulation to non-governmental body

Response:

CSOAI complements, not replaces national authority
Countries retain ultimate sovereignty
Can opt out of mutual recognition
Democratic governance (countries can participate in CSOAI)

19.9.3 Resource Constraints

Challenge: Not all countries can build AI expertise

Solution:

CSOAI provides shared capacity
Technical assistance to developing countries
Training programs for national regulators
South-South cooperation

19.10 CONCLUSION

CSOAI integrates with—not replaces—global regulatory frameworks. Through mutual recognition, coordinated standards, and shared governance, we create regulatory interoperability.

One AI system. Multiple jurisdictions. Single comprehensive assessment.

CSOAI + EU AI Act + NIST RMF + ISO Standards = Complete Compliance

Benefits:

Companies: Efficiency, clarity, reduced burden
Regulators: Shared expertise, coordinated action
Public: Consistent protection globally

This is how we govern AI in an interconnected world.

Effective Date: January 15, 2026, 09:00 GMT "One Planet, One Safety Framework, Many Voices"

REFERENCES

European Commission. (2021). Proposal for a Regulation on Artificial Intelligence (AI Act). COM(2021) 206 final.

NIST. (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1.

ISO/IEC. (2023). ISO/IEC 42001:2023 - Artificial Intelligence Management System.

OECD. (2019). OECD Principles on Artificial Intelligence. OECD/LEGAL/0449.

UNESCO. (2021). Recommendation on the Ethics of Artificial Intelligence.

UK Government. (2023). AI Regulation: A Pro-Innovation Approach. White Paper.

END OF ARTICLE 19

Next: Article 20 - Technical Standards & Specifications

From charter to certificate. This article is part of the standard behind Watchdog Certification — independent assessment, Ed25519-signed, publicly verifiable. The crosswalks to the EU AI Act, ISO/IEC 42001 and 18 more frameworks are in the Crosswalk Library; the runtime tools are in the fabric.

The 52-Article Charter is published in full in the Journal. Bespoke briefings: hello@meok.ai.