Skip to content

International Coordination: Research Report

📋Page Status
Quality:3 (Stub)⚠️
Words:1.1k
Backlinks:9
Structure:
📊 15📈 0🔗 4📚 5•3%Score: 11/15
FindingKey DataImplication
No binding treatiesZero AI-specific treatiesNo hard constraints
Soft law growingBletchley, Seoul declarationsNorms developing
US-China divideVery limited cooperationMajor obstacle
AI Safety Institutes10+ countries establishingTechnical coordination
Timeline pressureAI faster than diplomacyGovernance lag

International coordination on AI governance is essential because AI development is global, AI systems cross borders easily, and AI risks—from misalignment to misuse—affect all of humanity. Yet international coordination remains weak. No binding international treaties govern AI development. The major AI powers—the United States and China—are in strategic competition that makes cooperation difficult. And the pace of AI development far exceeds the pace of international diplomacy.

Recent efforts have made progress. The November 2023 Bletchley Declaration brought 28 countries together on AI safety principles. The May 2024 Seoul Summit expanded this with safety commitments from frontier AI labs. AI Safety Institutes are being established in multiple countries, creating potential for technical coordination. The G7 and G20 have addressed AI. But these remain voluntary frameworks without enforcement mechanisms.

The fundamental challenge is that international AI coordination requires cooperation between competitors. The US and China see AI leadership as essential to national security and economic competitiveness. Neither wants to constrain their own development or give the other an advantage. This creates a coordination problem where both might prefer mutual restraint but neither will move first.


MechanismDescriptionAI Status
TreatiesBinding international lawNone on AI
Soft lawNon-binding agreementsGrowing
StandardsTechnical specificationsDeveloping
InstitutionsInternational bodiesLimited AI mandate
BilateralCountry-to-countryVery limited
DomainTime to Major TreatyLessons
Nuclear20+ years (NPT 1968)Took near-disaster
Chemical weapons100+ years (CWC 1993)Very slow
Climate30+ years (Paris 2015)Still inadequate
CyberNo binding treaty yetVery difficult
AITBDMay be harder

MechanismParticipantsStatusEnforcement
Bletchley Declaration28 countriesActiveNone
Seoul Declaration27 countries, 16 companiesActiveVoluntary
G7 AI Code of ConductG7 countriesActiveVoluntary
GPAI29 countriesActiveNone
AI Safety Institutes network10+ countriesBuildingTechnical cooperation
DimensionCooperation LevelObstacle
Safety researchVery limitedCompetition concerns
StandardsCompeting standardsStrategic interests
Military AINoneSecurity concerns
Commercial AIMinimalTrade tensions
DialogueSporadicPolitical tensions
RegionApproachStatus
EURegulatory leadershipAI Act implemented
USLight touch + export controlsEO + agency rules
ChinaState-directedMultiple regulations
UKSafety research + bridge roleAI Safety Institute
OthersFollowing leadersVarious
InstitutionAI MandateEffectiveness
UNLimitedSlow, politicized
ITUTechnical standardsLimited AI role
OECDPolicy guidanceSoft law
ISOTechnical standardsDeveloping
Custom bodyNone yetProposed

FactorMechanismSeverity
Strategic competitionUS-China rivalryHigh
Sovereignty concernsStates resist constraintsHigh
Pace mismatchAI faster than diplomacyHigh
Verification difficultyCan’t verify AI complianceHigh
Capability diffusionMany actors, hard to coordinateModerate
FactorMechanismStatus
Shared risk perceptionCommon threat motivatesGrowing
Technical communityResearchers collaborateActive
Safety incidentsCrisis creates willNot yet occurred
Economic interdependenceCostly to decoupleWeakening
Norm entrepreneurshipCountries lead on safetyUK, others trying

ElementDescriptionFeasibility
Information sharingShare safety researchModerate
Hot linesCrisis communicationFeasible
Confidence buildingReduce misperceptionFeasible
Technical standardsInteroperabilityModerate
ElementDescriptionFeasibility
Compute monitoringTrack large training runsProposed
Incident reportingShare failuresPossible
Evaluation standardsCommon safety testsDeveloping
Mutual recognitionAccept others’ certificationsPossible
ElementDescriptionFeasibility
Binding treatyEnforceable limitsVery difficult
International bodyAI governance institutionProposed
Compute governanceControl training resourcesVery difficult
Racing limitsMutual capability constraintsVery difficult

ImplicationDescription
Racing continuesNo constraints on competition
Governance gapsSome jurisdictions unregulated
Lowest common denominatorStandards may race down
Incident responseNo coordinated response
ImplicationDescription
National focusMost governance will be national
Coalition of willingLike-minded countries coordinate
Technical trackTechnical cooperation easier than political
Long timelineMajor coordination may take decades

Related ParameterConnection
Societal TrustInternational trust enables coordination
GovernanceInternational is part of governance
Coordination CapacitySpecific to international level
Regulatory CapacityInternational bodies have limited capacity